Hippocampal theta coordinates memory processing during visual exploration

James E Kragel; Stephen VanHaerents; Jessica W Templer; Stephan Schuele; Joshua M Rosenow; Aneesha S Nilakantan; Donna J Bridge

doi:10.7554/eLife.52108

. 2020 Mar 13;9:e52108. doi: 10.7554/eLife.52108

Hippocampal theta coordinates memory processing during visual exploration

James E Kragel ^1,^✉, Stephen VanHaerents ², Jessica W Templer ², Stephan Schuele ², Joshua M Rosenow ³, Aneesha S Nilakantan ¹, Donna J Bridge ^1,^✉

Editors: Laura L Colgin⁴, Laura L Colgin⁵

PMCID: PMC7069726 PMID: 32167468

Abstract

The hippocampus supports memory encoding and retrieval, which may occur at distinct phases of the theta cycle. These processes dynamically interact over rapid timescales, especially when sensory information conflicts with memory. The ability to link hippocampal dynamics to memory-guided behaviors has been limited by experiments that lack the temporal resolution to segregate encoding and retrieval. Here, we simultaneously tracked eye movements and hippocampal field potentials while neurosurgical patients performed a spatial memory task. Phase-locking at the peak of theta preceded fixations to retrieved locations, indicating that the hippocampus coordinates memory-guided eye movements. In contrast, phase-locking at the trough of theta followed fixations to novel object-locations and predicted intact memory of the original location. Theta-gamma phase amplitude coupling increased during fixations to conflicting visual content, but predicted memory updating. Hippocampal theta thus supports learning through two interleaved processes: strengthening encoding of novel information and guiding exploration based on prior experience.

Research organism: Human

Introduction

Hippocampal theta rhythms are prominent during active exploration of novel environments, perhaps due to encoding and retrieval processes necessary to guide ongoing behavior. Interactions between encoding and retrieval support many important functions, including memory updating, which requires comparing novel sensory inputs to prior memories and integrating new content into memory representations (Bridge and Paller, 2012; Bridge and Voss, 2014b). Retrieval-mediated reconsolidation requires the presence of novel information during reactivation, suggesting that this hippocampal-dependent learning process is sensitive to mismatch (associative novelty) between the retrieved content and sensory input (Morris et al., 2006; Winters et al., 2011). Some studies have even demonstrated hippocampal involvement in associative novelty (Bridge and Voss, 2014a; Chen et al., 2013; Duncan et al., 2009; Duncan et al., 2012; Honey et al., 1998; Howard et al., 2011; Kumaran and Maguire, 2007a; Kumaran and Maguire, 2009; Long et al., 2016; Thakral et al., 2015). However, the underlying novelty and retrieval processes have not been segregated as they unfold, in part because it is difficult to segregate these mechanisms in real time, as they interact continuously during learning. Many experimental designs capitalize on an artificial separation between encoding and retrieval phases, but these designs do not capture the natural interplay between these states that guides exploratory behavior and informs decision-making. Here, we assayed the engagement of encoding and retrieval processing in real time by designing a task to link memory-guided eye movements to intracranial recordings of hippocampal activity, and aimed to identify how theta oscillations are distinctly involved in encoding and retrieval processes.

Eye movements provide rich temporal information regarding the focus of attention and the specific cognitive processes engaged at any given moment (Bridge et al., 2017; Bridge and Voss, 2014b; Bridge and Voss, 2015; Voss et al., 2011). In human and non-human primates, learning through exploration heavily depends on the visual system (Meister and Buffalo, 2016), with eye movements resetting the phase of theta during learning of novel visual information (Hoffman et al., 2013; Jutras et al., 2013). Eye movements are also deployed rapidly, with median saccade rates of 3–5 Hz during visual exploration tasks (Wilming et al., 2017). Thus, eye movements provide an ideal behavioral measure to dissect learning processes on a moment-to-moment basis.

The interaction of fast gamma and theta oscillations in the hippocampus could play a key role in coordinating interactions between encoding and retrieval during exploration. The amplitude of gamma increases at specific phases of the theta cycle to support memory processing (Axmacher et al., 2010; Lega et al., 2016). Computational models of hippocampal function suggest that gamma activity associated with encoding and retrieval preferentially occurs at the trough and peak of the theta rhythm, respectively (Hasselmo et al., 2002). Task-based observations of hippocampal firing support this idea, showing that novel stimulus encoding and memory retrieval are enhanced at distinct phases of the theta cycle (Douchamps et al., 2013; Lever et al., 2010; Manns et al., 2007; Newman et al., 2013). In addition, closed-loop optogenetic stimulation of inhibitory neurons aligned to the peak of theta improves encoding, whereas stimulation aligned to the trough improves retrieval (Siegle and Wilson, 2014).

In rodents, theta-modulated gamma band activity has also distinguished encoding from retrieval (Bragin et al., 1995; Colgin, 2015). Distinct slow (~30 to 50 Hz) and mid (~50 Hz to 100 Hz) gamma oscillations are observed and generated by separate neural circuits, with maximal amplitudes at the peak and trough of the theta rhythm, respectively (Colgin, 2016). Recent work has also identified unique theta-nested gamma oscillations observed within individual theta cycles (Lopes-Dos-Santos et al., 2018), providing support for the notion that fast and slow gamma separately mediate encoding of novel information and memory retrieval. Intracranial recordings in humans have identified ripple oscillations (~80 to 100 Hz) that are involved in memory retrieval and consolidation (Axmacher et al., 2008; Staresina et al., 2015; Vaz et al., 2019) and exhibit phase amplitude coupling (PAC) with hippocampal theta phase (Staresina et al., 2015). Similar ripple oscillations are also prevalent in nonhuman primates during visual search (Leonard and Hoffman, 2017; Leonard et al., 2015), raising the possibility that they play an active role in exploration. Because evidence for memory-related hippocampal theta to gamma PAC in humans has primarily focused on verbal learning tasks (Lega et al., 2016; Mormann et al., 2005), it is not known how changes in PAC during exploration translate from animal models to similar behaviors in humans.

Here, we recorded eye movements and intracranial hippocampal recordings as neurosurgical patients performed an associative spatial memory task. We hypothesized that theta oscillations would influence when eye movements were driven by either associative novelty or memory retrieval, through theta-dependent modulation of neuronal firing. Indeed, previous work in humans has examined the relation between theta oscillations and memory in verbal recall tasks (Kahana et al., 2001; Lega et al., 2012; Sederberg et al., 2003), including theta-gamma phase amplitude coupling (Lega et al., 2016; Mormann et al., 2005; Vaz et al., 2017). These studies examined encoding and retrieval in isolated task epochs, raising the question of how theta supports these memory processes when they rapidly co-occur. In our spatial memory task, subjects encountered previously studied objects in either their original or updated spatial locations. When an object appeared in an updated location, subjects directed viewing to both the updated and original locations iteratively over the course of the trial. Simultaneous acquisition of eye movements and intracranial EEG allowed us to relate theta to the timing of encoding and retrieval processes with exceptional temporal resolution. In doing so, we systematically tested the hypothesis that hippocampal theta influences when different memory processes occur during exploratory viewing. If the strength of encoding and retrieval are modulated by theta phase, fixations driven by each process should be phase-locked to the theta rhythm. In addition, eye-movements tied to distinct encoding and retrieval processes should occur at distinct phases of theta with variation in hippocampal PAC. As such, this experiment determines how the hippocampus contributes to learning and coordinates dynamic encoding and retrieval operations during visual exploration in humans.

Results

Direct brain recordings linked to memory-guided eye movements

Subjects performed a multi-phase associative spatial memory task (Figure 1a), while we simultaneously recorded eye movements and local field potentials from the hippocampus (Figure 1b). During the study phase, subjects learned the spatial location of 16 objects presented sequentially on a background scene. Next, during a refresh phase, objects were re-presented in either repeated (Match) or updated (Mismatch) spatial locations, with two visual cues (small red dots) indicating potential alternate locations. One visual cue always indicated the object’s original location during Mismatch trials. After viewing each stimulus, subjects indicated via button press whether each object was presented in its original or updated location. During a final recognition phase, subjects viewed each object in three locations and attempted to identify the object’s original location. All subjects completed eight blocks of the study-refresh-recognition sequence, with 16 unique objects per block (128 total), and a unique background scene per block (eight total).

During the refresh phase, subjects were encouraged to visually explore the three cued locations to help inform their memory decision. Our primary analyses focused on the interplay of associative novelty and retrieval processes during these Mismatch trials, by linking hippocampal activity to eye movements directed to the original and updated locations. We use the term retrieval-dependent to refer to fixations to the original object-location on Mismatch trials, as retrieval of spatial information is necessary for preferential viewing of this location. Novelty-dependent fixations are driven to updated object-locations, as compared to fixations to objects presented in repeated locations during Match trials. By leveraging eye movements in this manner, we were able to identify distinct hippocampal mechanisms linked to these cognitive processes. In addition, we evaluated the impact of viewing behaviors and electrophysiological states on final recognition performance.

We measured overall task performance by computing accuracy on the final recognition test. Subjects performed the task well, correctly identifying repeated object-locations on 72% (±6 SEM) of Match trials and novel object-locations on 70% (±6 SEM) of Mismatch trials during the refresh phase. Subjects remembered the original object-location on 72% (±6 SEM) of final recognition test trials (Figure 1c). We assessed how the factors of condition (Mismatch/Match) and refresh task performance (Correct/Incorrect) influenced memory for the original object-location using a two-factor repeated measures ANOVA. We observed significant main effects of refresh performance (F_1,4 = 115.0, p = 0.0004, η_p² = 0.97) and condition (F_1,4 = 13.4, p = 0.02, η_p² = 0.77), without evidence for an interaction (F_1,4 = 0.4, p = 0.55, η_p² = 0.10). Recognition accuracy on the final test was significantly better following correct (M = 84, SD = 10) than incorrect (M = 35, SD = 15) judgments on the refresh phase (paired t-test, t₄ = 10.72, p = 0.0004, g = 3.4). In addition, accuracy on the final recognition test was significantly impaired on Mismatch (M = 63, SD = 14) relative to Match trials (M = 82, SD = 14; paired t-test, t₄ = 9.83, p = 0.0006, g = 1.2).

To confirm that eye movements during the refresh phase were tied to memory processes, we examined changes in viewing behaviors and memory outcomes on the final recognition test (Table 1). On average, participants made more fixations to the presented object during Match (M = 5.3, SD = 1.0; paired t-test, t₄ = 12.9, p = 0.0002, g = 2.9) and Mismatch (M = 4.6, SD = 0.9; paired t-test, t₄ = 17.7, p = 0.0001, g = 2.8) trials than to the other two cued locations on Mismatch (M = 2.4, SD = 0.8) and Match (M = 1.7, SD = 0.8) trials. Notably, the number of fixations to the object was reduced on Mismatch relative to Match trials (paired t-test, t₄ = −8.8, p = 0.0009, g = −0.6), indicating increased exploration during Mismatch. In addition, fixation durations to objects were longer than matched spatial cues (paired t-test, all t₄ > 3.4, p < 0.03, g > 1.8) and were comparable between Match (M = 342 ms, SD = 88) and Mismatch trials (M = 388, SD = 131; paired t-test, t₄ = 1.49, p = 0.21, g = 0.4).

Table 1. Task-related eye movement behavior.

		Mismatch			Match
	Original	Updated	Other	Repeated	Other	Other
Fixations per trial (N)	2.4 (0.3)	4.6 (0.4)	1.4 (0.3)	5.3 (0.4)	1.7 (0.4)	1.4 (0.3)
Fixation SME (t)	1.4	−4.6^*	−0.3	3.7^*	−0.4	−1.7
Fixation duration (ms)	210 (11)	388 (59)	212 (14)	342 (39)	199 (11)	206 (6)
Duration SME (t)	0.5	1.1	1.4	−1.6	−2.4	−0.6

Open in a new tab

Group-level description of eye movement behavior to six viewing regions of interest. The subsequent memory effect (SME) for each measure was assessed by a one-sample t-test, across subjects (n = 5). ^*P < 0.05. Parentheses denote standard error of the mean.

Viewing behavior on Match and Mismatch trials predicted final recognition performance. On Match trials, the number of fixations to the repeated object predicted better memory for the original location, whereas the number of fixations to the updated location on Mismatch trials predicted memory updating (see Table 1 for details). To break down the timing of these memory-guided eye movements, we examined the proportion of time spent viewing each region of interest (ROI) across trials (Figure 1d). We found that viewing preferences on Mismatch trials, but not Match trials, predicted later memory for the original object-locations. Following initial visual orienting to the novel stimulus, prolonged viewing of the novel object-location (738 to 2188 ms after object presentation) was associated with memory updating (nonparametric cluster test, P_FWE < 0.05, g = −0.7, n = 5 subjects). Viewing the original object-location during this time period (from 998 to 1584 ms) led to better memory for the original location (nonparametric cluster test, P_FWE < 0.05, g = 1.2, n = 5 subjects). Proportion of viewing over time during Match trials was not a significant predictor of final memory performance (P_FWE > 0.05). These findings suggest that interplay between memory processes and visual sampling during Mismatch trials determined whether memory updating would occur.

Theta dependence of memory-guided eye movements

We analyzed direct recordings from hippocampal depth electrodes in five subjects (Figure 2a), referencing signals from contacts in hippocampus or adjacent white matter to bipolar pairs (hereafter called electrodes). Average power spectra from pre- and post-fixation intervals contained theta peaks irrespective of task condition and fixation target (Figure 2b). To determine the consistency of these peaks at the individual electrode level (see Figure 2—figure supplement 1 for individual power spectra), we modeled aperiodic and oscillatory components of power spectra for individual fixations. We tested whether theta (4 to 6 Hz) oscillations were present by comparing variability in the full spectra (which includes oscillatory components) to the amount explained by aperiodic components alone. A majority of electrodes exhibited theta oscillations around this peak frequency both before (−750 to 50 ms) and after (−50 to 750 ms) fixation onsets to each ROI (see Table 2 for details). In the following analyses, we focused on spectral power and phase from 1 Hz to 10 Hz, which includes both theta and low-theta (Jacobs, 2014; Watrous et al., 2013a) frequency bands previously associated with visual exploration and memory encoding (Jutras et al., 2013; Lega et al., 2012). We linked measures of spectral power and phase to individual fixation events to identify hippocampal states reflecting retrieval and novelty detection. We reasoned that hippocampal signaling prior to fixations would reflect a memory-guided initiation of the upcoming eye movement, whereas signaling following fixations would reflect a memory-based reaction to visual input.

Figure 2—figure supplement 1. — (a) Location of hippocampal electrodes in MNI space. (b) Mean power spectra in peri-fixation epochs. Spectral peaks in the 4 to 6 Hz range are shown in epochs preceding and following fixations to all regions of interest. (c) Increased phase-locking precedes retrieval-dependent fixations. Significant differences (cluster P_FWE < 0.05) in inter-trial phase clustering (ITC) between fixations (indicated by the dashed line) to original vs. updated object-locations on Mismatch trials are highlighted. (d) Novelty related modulations in hippocampal phase. Significant differences in ITC following fixations to the updated object-location on Mismatch trials and the repeated object-location on Match trials.

Figure 2—source data 1. MATLAB code and source files to reproduce data in Figure 2.

elife-52108-fig2-data1.zip^{(1.2MB, zip)}

Table 2. Proportion of electrodes showing theta (4 to 6 Hz) oscillations.

			Pre fixation			Post fixation
	N	Original	Updated	Repeated	Original	Updated	Repeated
S1	8	0.75^*	0.75^*	0.38^*	0.25	0.5^*	0.5^*
S2	4	0	0.25	0.75^*	0	0.5	0.75^*
S3	6	1.0^*	1.0^*	1.0^*	0.67^*	1.0^*	1.0^*
S4	8	1.0^*	1.0^*	1.0^*	1.0^*	1.0^*	1.0^*
S5	6	0.83^*	0.83^*	0.83^*	0.5^*	0.67^*	0.67^*
Group	32	0.78^*	0.81^*	0.78^*	0.53^*	0.75^*	0.78^*

Open in a new tab

Proportion of electrodes showing significant theta oscillations, before (Pre, −750 to 50 ms) and after (Post, −50 to 750 ms) fixations to locations of interest. p-Values were computed using a binomial test, based on the total number of electrodes (N) for a subject or group. ^*p < 0.05, Bonferroni corrected across locations and time intervals.

Theta phase coherence increases during retrieval and novelty detection

To assess whether retrieval-dependent eye movements during Mismatch trials occurred at specific phases of hippocampal oscillations, we compared the consistency of phase angles in the moments leading up to fixations to the original and updated locations using inter-trial phase coherence (ITC; Figure 2c). We observed significantly greater phase-locking around 5 Hz prior to fixations to the original compared to the updated object-location on Mismatch trials (nonparametric cluster test, P_FWE = 0.045, g = 1.2, n = 5 subjects). In an additional analysis, we also identified significantly greater phase-locking preceding fixations to the original object-location relative to fixations to objects presented in repeated locations (Match condition) during the same time-frequency window (Figure 2—figure supplement 2, P_FWE = 0.04, g = 3.4, n = 5 subjects). Theta power was comparable before fixations to the original and updated object-locations (one sample t-test, t₄ = −1.6, p = 0.17, Figure 2—figure supplement 3), indicating that differences in ITC did not result from reductions in theta magnitude.

As fixations to the original object-location were frequently preceded by novelty-driven fixations (Figure 1d), it is possible that the observed phase-locking effect resulted from novelty detection rather than retrieval. Two control analyses suggested this was not the case. First, ITC measured during fixations to the updated location did not differ depending on the target of the next saccade (i.e. either to the original or updated location; all P_FWE > 0.15, n = 5 subjects). Second, we observed significantly increased theta phase-locking across subjects (one sample t-test, t₄ = 2.95, p = 0.04) during the same time interval when we excluded fixations that were preceded by fixations to the updated object-location (which could cause stimulus-related processing to occur prior to the retrieval-guided fixation). These findings suggest that the observed pre-fixation theta effects reflect a retrieval mechanism, rather than novelty-related processes that initiate memory retrieval.

We next asked whether theta phase was modulated following fixation onset. If theta phase is generally modulated by fixations during memory updating (i.e. during Mismatch trials), consistent phase-locking would occur irrespective of the viewing location. To test this possibility, we contrasted ITC between each type of fixation (to the original or updated location) on Mismatch trials with fixations to the repeated location on Match trials. We restricted this analysis to trials where memory for the original object-location was intact, to increase the likelihood that fixations were driven by associative novelty rather than purely stimulus-driven factors. Significantly (nonparametric cluster test, P_FWE < 0.001, g = 1.2, n = 5 subjects) greater phase clustering followed fixations to updated compared to repeated object-locations (Figure 2d). This post-saccade phase-locking effect only followed fixations to updated locations, as ITC did not significantly differ between fixations to the Mismatch-original and Match-repeated locations. In addition, we did not find any significant differences between phase-locking following fixations to original and updated locations during Mismatch trials. Taken together, our comparison of ITC between fixation targets suggests modulation of theta phase by two processes: retrieval processing preceding fixations to the original location and novelty processing following fixations to the updated location.

Theta phase angle differentiates retrieval and novelty detection

Models of memory-related oscillations propose that encoding and retrieval operations preferentially occur during the trough (π rad) or peak (0 rad) of hippocampal theta, respectively (Hasselmo et al., 2002). If theta phase-locking preceding fixations to original object-locations was driven by theta-dependent retrieval mechanisms, we would predict preferred phase angles near the peak of the theta. On the other hand, we would predict the phase-locking to begin at the trough of theta oscillations during fixations to updated object-locations, as theta troughs are associated with increased sensory inputs from the entorhinal cortex that support encoding of the environment.

To test these predictions, we examined differences in theta phase when encoding and retrieval processing occurred on Mismatch trials. The time course of theta coherence (Figure 3a) reveals the timing of hippocampal retrieval and novelty processes; a pre-fixation retrieval effect and a reset of theta phase due to fixations to updated object-locations. We compared phase angles during two time periods of increased ITC: before fixations to the original location (Pre, −420 to −380 ms) and following fixations to the original location (Post, 80 to 120 ms) on Mismatch trials (for an example trial, see Figure 3b). As absolute phase angles can be difficult to interpret due to phase reversals caused by referencing (Shirhatti et al., 2016), we evaluated whether differences in phase between these time intervals were consistent for individual electrodes. We found significant differences (p < 0.05, uncorrected, permutation test) in 27 out of 32 electrodes, significantly more than expected by chance (binomial test, p <0.0001). Phase distributions for one example electrode are depicted in Figure 3c, showing peak-concentrated phase preceding fixations to the original object-location (mean = −0.40 rad, 95% CI [0.69–1.49], n = 179 fixations), and trough-concentrated phase following fixations to the updated location (mean = 2.85 rad, 95% CI [2.51 3.19], n = 244 fixations). Indeed, the average phase of these two distributions were significantly different (Z_F = 10.3, p < 0.0001, permutation test).

We next tested whether preferred phases were consistent across recording sites (Figure 3d). Average theta phase preceding fixations to the original object-location occurred near the peak of theta (mean = 0.2 rad, 95% CI [−0.7 1.0]). In contrast, the average theta phase following fixations to the updated object-location occurred near the trough of theta (mean = 2.8 rad, 95% CI [2.3 3.3]). Permutation testing revealed consistent differences in phase across electrodes (Z_F = 6.8, permutation p < 0.001, n = 32 electrodes). While these phase differences between conditions appear to be driven by memory processing, it is possible that they resulted from differences in timing alone. Two pieces of evidence refute this conclusion. First, the distribution of phase in the interval preceding fixations to updated object-locations is indistinguishable from a uniform distribution (Z = 1.04, p = 0.36, Rayleigh test, n = 32 electrodes). Second, following fixations to the original object-location, we found phase concentrated around theta troughs (mean = 2.9 rad, 95% CI [2.5 3.4]). These phases were not distinguishable from those following fixations to the updated object-location (F₆₂ = 0.77, p = 0.38, n = 32 electrodes). These results provide additional evidence that the timing of memory-dependent fixations depend upon the phase of theta.

Theta phase consistency during updated-location viewing predicts subsequent memory

If the observed phase-locking of eye movements to hippocampal theta reflects encoding and retrieval operations, we would predict that the strength of coherence between visual sampling and theta determines memory performance. To test this prediction, we tested whether ITC predicted subsequent memory performance (i.e. differed between trials where the original location was subsequently remembered or forgotten) on the final recognition test (Figure 4). We observed no significant subsequent memory effects following fixations to the original object-location (Figure 4a, nonparametric cluster test, P_FWE = 0.38, n = 5 subjects). Conversely, phase-locking following fixations to updated locations during Mismatch trials predicted subsequent memory performance (Figure 4b). Greater ITC at frequencies ranging from 3 to 6 Hz following fixations to the updated location was associated with better memory for the original location on the final recognition test (nonparametric cluster test, P_FWE < 0.0001, g = 0.5, n = 5 subjects). The ability of hippocampal phase-locking to predict memory outcomes was specific to viewing updated object-locations; this subsequent memory effect was significantly greater following fixations to updated as opposed to original object-locations (Figure 4c, nonparametric cluster test, P_FWE < 0.0001, n = 5 subjects). Furthermore, these differences could not be accounted for by differences in power (all P_FWE > 0.05, nonparametric cluster correction, n = 5 subjects). These results suggest that the reset of hippocampal theta by novel sensory information may be a major determinant of memory performance on this task. Moreover, the pattern of viewing behavior on Mismatch trials indicates that the detection of the updated object-location precedes retrieval-guided viewing of the original object-location (Figure 1d), suggesting that retrieval-related changes in ITC were predictive of memory-guided visual exploration but not memory outcomes.

Time-frequency plots depict differences in inter-trial phase coherence (ITC) between subsequently remembered and forgotten Mismatch trials. (a) No subsequent memory effects were present during fixations to the original object-location. (b) Significant (P_FWE < 0.05, nonparametric cluster corrected) increases in phase-locking were associated with memory following the fixations to the updated object-location. (c) Subsequent memory effects were specific to updated object-locations, as revealed by the significant (P_FWE < 0.05, nonparametric cluster corrected) interaction following fixation onset.

Figure 4—source data 1. MATLAB code and source files to reproduce data in Figure 4.

elife-52108-fig4-data1.zip^{(1.4MB, zip)}

Figure 4—figure supplement 1. — Time-frequency plots depict differences in inter-trial phase coherence (ITC) between subsequently remembered and forgotten Mismatch trials. (a) No subsequent memory effects were present during fixations to the original object-location. (b) Significant (P_FWE < 0.05, nonparametric cluster corrected) increases in phase-locking were associated with memory following the fixations to the updated object-location. (c) Subsequent memory effects were specific to updated object-locations, as revealed by the significant (P_FWE < 0.05, nonparametric cluster corrected) interaction following fixation onset.

Figure 4—source data 1. MATLAB code and source files to reproduce data in Figure 4.

elife-52108-fig4-data1.zip^{(1.4MB, zip)}

Gaze-dependent theta modulations are consistent across subjects and electrodes

In the previously reported analyses of hippocampal phase and memory processes, we adopted the conservative approach of group-level inference (i.e. we tested for differences in the group mean while treating subject as a random factor), without focusing on variability in the observed effects across subjects or anatomical locations within the hippocampus. To supplement these findings, we report the electrode-level results of the main theta phase-locking analyses broken down by individual subjects (Table 3). While the magnitude of these effects is biased due to selection from significant group-level effects, we observed significant retrieval effects in about one third of all electrodes, and more than half showed significant novelty and subsequent memory effects. As such, the presented group results are common throughout our hippocampal recordings.

Table 3. Proportion of electrodes showing significant theta phase-locking effects.

	N	Retrieval	Associative novelty	Subsequent memory
S1	8	0.13	0.5^*	0.63^*
S2	4	0	0.75^*	0.5^*
S3	6	0.5^*	0.5^*	0.33
S4	8	0.5^*	0.75^*	0.75^*
S5	6	0.5^*	0.33	0.33
Group	32	0.34^*	0.56^*	0.53^*

Open in a new tab

Proportion of electrodes showing significant phase-locking effects. p-Values were computed using a binomial test, based on the total number of electrodes (N) for a subject or group. ^*p < 0.05, Bonferroni corrected across three conditions.

Rapid sequential fixations do not account for memory-specific theta effects

Our ability to associate hippocampal theta with specific memory processes assumes a strong correspondence between theta phase dynamics and the moments surrounding a fixation of interest. However, eye movements made during task performance were necessarily sequential in nature, as subjects iteratively viewed multiple locations on the screen. Thus, it is possible that the observed effects were influenced by these iterative behaviors (e.g. repeated viewing of novel locations). Indeed, subjects made multiple fixations during the epochs in which we observed significant phase-locking. To determine whether the observed effects were truly related to the fixations of interest, rather than preceding or following eye movements, we repeated our phase-locking analyses and restricted analysis to fixation events with a clean pre- or post-fixation window of 100, 200, and 400 ms in duration. While longer fixation-free windows greatly reduced our power to detect effects (see Supplementary file 1 for details on the number of eye movements contributing to each analysis), we replicated our major findings in these analyses. Notably, we observed significantly increased ITC preceding fixations to original vs. updated locations on Mismatch trials (Figure 4—figure supplement 1a). In addition, we observed similar increases in ITC following fixations to updated vs. repeated objects (Figure 4—figure supplement 1b), the magnitude of which predicted whether memory for the original object-location was maintained (Figure 4—figure supplement 1c). These results indicate that the observed phase effects are likely related to fixations of interest, rather than sequential behaviors.

Reset of hippocampal oscillations during memory-guided eye movements is specific to theta

While our analysis of phase dynamics focused on theta frequencies, it is possible that broadband or higher frequency (e.g. beta and gamma activity) phase dynamics are related to memory updating and retrieval. We tested for the presence of phase resets caused by individual fixations. From average ERPs at each channel, we computed pre-fixation (−750 to 50 ms) and post-fixation (−50 ms to 750 ms) power. A phase reset at a given frequency band would be indicated by an increase in post- vs. pre-fixation power, as phase alignment in this period would lead to increased power in the average ERP. We found evidence for a reset of ongoing oscillations in the 8 to 13 Hz range following fixations to presented objects, irrespective of condition (Figure 4—figure supplement 2). We observed significant differences (cluster P_FWE = 0.006, t₄ = −4.04, g = −1.14) in the magnitude of this phase reset effect when comparing fixations to the original and updated object-location on Mismatch trials (Figure 4—figure supplement 2a). Moreover, averaged ERPs locked to fixations to the original object-location exhibited increased power in the pre- vs. post-fixation interval at frequencies below 8 Hz (Figure 4—figure supplement 2a, cluster P_FWE = 0.004, t₄ = −3.7, g = −1.6). The magnitude of this effect was marginally different from fixations to updated locations (cluster P_FWE = 0.06, t₄ = −1.5, g = −0.6). These findings provide additional support for coordination between theta phase and the deployment of retrieval-dependent eye movements, as phase consistency increased prior to these retrieval-guided fixations. Finally, phase resetting following fixations to the updated object-location on Mismatch trials (Figure 4—figure supplement 2c) marginally predicted memory for the original object-location (cluster P_FWE = 0.09, t₄ = 2.1, g = 1.08). We did not observe evidence for phase resets at frequencies above 13 Hz, near the upper border of the alpha band.

Theta to gamma phase amplitude coupling predicts memory updating

Having identified a consistent relationship between theta phase and specific viewing behaviors, we examined the relationship between theta phase and the amplitude of high frequency (80–200 Hz) gamma band activity. Phase-amplitude coupling (PAC) between theta and gamma has been proposed as a mechanism for separating memory representations (Hasselmo and Eichenbaum, 2005), with supporting evidence in both animal models (Tort et al., 2009) and humans (Axmacher et al., 2010; Heusser et al., 2016; Lega et al., 2016; Lisman and Jensen, 2013; Vaz et al., 2017). We used the modulation index (MI) to quantify PAC between the phase of theta (ranging from 1 to 10 Hz) and gamma amplitude (Tort et al., 2010).

We focused our PAC analyses on three comparisons of interest: associative novelty, memory retrieval, and memory performance. Results from an example electrode are depicted in Figure 5, showing increases in PAC related to forgetting of the original object-location. For a given electrode (Figure 5a), we computed gamma amplitude as a function of theta phase during each trial type of interest. These measures were used to compute the MI, which measures PAC as the difference between the observed amplitude distribution and a uniform distribution (Figure 5b, dashed line). To make sure observed differences in PAC did not result from non-stationarities in the data or common task-evoked changes in amplitude and phase (Aru et al., 2015), we permuted phase information across trials for each condition and computed a normalized score (MI_Z) based on this null distribution. We assessed changes in MI_Z across conditions (Figure 5c) to identify memory-related changes in PAC. Across subjects, we found 25% of electrodes exhibited significant differences in PAC driven by associative novelty (i.e. differences between fixations to updated and repeated locations), significantly more than expected by chance (binomial test, p < 0.001, proportion difference = 0.2). This effect was primarily driven by increased PAC during the viewing of updated locations. We found significantly greater PAC on 16% of electrodes (P_FWE < 0.05, permutation test), compared to surrogate measures obtained by permuting theta phase across fixation events (significantly more than expected by chance, binomial test, p = 0.02). Only 9% of electrodes exhibited significant differences in PAC preceding fixations to the original versus updated locations on Mismatch trials (binomial test, p = 0.14, proportion difference = . 05).

(a) Re-referenced bipolar recording from contacts in the left hippocampus and adjacent white matter. (b) Normalized amplitude distributions reveal memory-related modulation of gamma (150 Hz) amplitude by theta (4 Hz) phase at this recording site. MI, modulation index. Dashed line denotes normalized gamma amplitude under a uniform distribution. (c) Left, comodulograms depict increased PAC (z-scored MI, constructed from trial-shuffled surrogate data) during fixations to updated object-locations when memory for the original object-location was forgotten. Right, the statistical map depicts a cluster of significant (P_FWE < 0.05, nonparametric cluster corrected) cross-frequency interactions.

Figure 5—source data 1. MATLAB code and source files to reproduce data in Figure 5.

elife-52108-fig5-data1.zip^{(75.9MB, zip)}

Next, we tested for group-level differences in PAC during specific viewing behaviors. To account for variability in theta frequency across subjects and electrodes, we selected the theta frequency that exhibited the greatest magnitude MI_Z from 4 to 6 Hz, irrespective of condition. We first examined if theta to gamma PAC was sensitive to associative novelty by contrasting PAC following fixations to updated versus repeated objects. We found significantly increased theta to gamma (80 to 100 Hz) PAC during fixations to updated object-locations (nonparametric cluster test, P_FWE = 0.05, g = 1.2, n = 5 subjects), indicating that gamma amplitude was more dependent on theta phase when visual stimuli conflicted with memory (Figure 6a).

(a) Top, post-saccade changes in PAC following (−50 ms to 750 ms) fixations to updated and repeated object-locations are displayed for a range of gamma amplitudes. Below, significant (P_FWE < 0.05, nonparametric cluster corrected) increases in PAC related to novelty are indicated. (b) Theta to gamma PAC during (−50 ms to 750 ms) fixations to updated object-locations varies with memory outcome. A significant (P_FWE < 0.05, nonparametric cluster corrected), negative subsequent memory effects is depicted in the bottom panel. (c) PAC did not differ in the moments leading up to (−750 ms to 50 ms) fixations to updated and original object-locations during Mismatch trials. List of Tables.

Figure 6—source data 1. MATLAB code and source files to reproduce data in Figure 6.

elife-52108-fig6-data1.zip^{(17.8MB, zip)}

Given significant PAC effects during fixations to updated object-locations, we next evaluated whether PAC during these fixations predicted subsequent memory performance. We found that increases in theta to high-gamma (130 to 150 Hz) PAC were significantly greater (nonparametric cluster test, P_FWE = 0.04, g = −0.9, n = 5 subjects) on trials where the original object-location was subsequently forgotten (Figure 6b). Because forgetting was caused by interference from updated object-locations (subjects chose the updated location on 73% (SD = 3%) of final recognition trials), these results indicate that increased theta to high-gamma PAC reflects updating of the object-location in memory. Finally, we tested whether PAC varied with retrieval demands by comparing measures of PAC in the moments preceding fixations to the original and updated object-locations on Mismatch trials (Figure 6c). We did not observe a consistent relationship between PAC and retrieval-related eye movements, consistent with our electrode-level analyses.

As we observed different gamma frequencies modulated by theta associated with novelty detection vs. subsequent memory, we aimed to compare the specificity of these effects directly. To do this, we compared the difference in MI_z between the two frequency ranges for each contrast. This analysis suggested specificity of memory updating in the high-gamma range (130–150 Hz), as it produced greater subsequent memory effects compared to the 80 to 100 Hz gamma band associated with novelty detection (paired t-test, t₄ = 2.6, p = 0.06, g = 0.7). On the other hand, we did not observe this level of specificity associated with associative novelty (paired t-test, t₄ = −0.2, p = 0.84, g = −0.1). These results implicate increased theta to high-gamma PAC in the updating of previously formed memory traces.

We examined the specificity of the observed PAC effects to different theta frequencies by repeating the aforementioned analyses for low-theta (1–3 Hz) and faster theta/alpha frequencies (7–10 Hz). We did not observe significant memory-related differences in PAC in these ranges. Directly comparing PAC effects across different ranges of theta, we found significantly greater associative novelty-related changes in PAC when using 4–6 Hz phase (identified via clear peaks in the power spectrum, see Figure 2b) to define the modulating frequency compared to the faster theta/alpha band (cluster P_FWE = 0.01, t₄ = 1.2, g = 0.7). There was weak evidence for stronger effects than the low-theta band (cluster P_FWE = 0.08, t₄ = 1.9, g = 1.3), suggesting specificity of associative novelty-related PAC in the 4–6 Hz theta range. No evidence for frequency specificity was found when comparing updating-related PAC effects across different ranges of theta (all cluster P_FWE > 0.13). The observed interactions between theta phase and gamma amplitude likely reflect distinct processing states in hippocampal networks, wherein asynchronous local activity is necessary to segregate novel perceptual information from previously encoded memories.

Memory-related changes in PAC are unrelated to theta waveform properties

Changes in theta to gamma PAC can arise from multiple sources such as the sharpness of non-sinusoidal oscillations, differences in oscillatory power, and phase-locking of ongoing oscillations to sensory events (Cole et al., 2017; Cole and Voytek, 2017; Vaz et al., 2017). We performed a series of control analyses to determine whether the observed statistical differences in PAC were likely to reflect a direct relationship between two distinct oscillatory features (i.e. modulation of gamma amplitude by theta phase) as opposed to other properties of theta oscillations. For each fixation of interest, we measured the average theta amplitude, sharpness of peaks and troughs, waveform asymmetry, and measures of phase-locking (the difference in phase angle between each trial and the average within-condition phase angle for each condition and time point). We then asked whether novelty- and updating-related changes in PAC (MI_Z) could be explained by changes in these properties of theta waveforms.

In our initial analysis of associative novelty (contrasting fixations to updated vs. repeated object-locations), we identified theta to gamma (80–100 Hz) PAC. We found that theta waveform properties were significantly related to PAC in this frequency range for a minimal number of electrodes (range 2 to 5 electrodes, see Table 4 for details), indicating that there was no consistent relation between theta waveform properties and novelty-related PAC. Theta waveform properties did not influence theta to gamma (130–150 Hz) PAC associated with memory updating, with one notable exception (see Table 5 for details). Changes in theta power predicted trial-level differences in PAC in 22% of electrodes, significantly more than expected by chance (p = 0.0008, binomial test).

Table 4. Proportion of electrodes showing changes in PAC due to theta waveform properties while viewing all objects during refresh.

	N	Power	Sin(θ)	S_peak	S_Trough	Asym	Full
S1	8	0	0.13	0	0.13	0	0.13
S2	4	0.50^*	0	0.25	0	0	0
S3	6	0	0.33	0	0	0	0.17
S4	8	0.25	0.13	0	0	0.25	0.13
S5	6	0.17	0.17	0.17	0.17	0.17	0.17
Group	32	0.16	0.16	0.06	0.06	0.09	0.13

Open in a new tab

Proportion of electrodes showing significant modulation of PAC by properties of theta, including power, single-trial measures of phase-locking, peak and trough sharpness, and waveform asymmetry. Each column denotes the parameters included in each regression. Model significance was determined with the F-statistic. ^*p < 0.05, Bonferroni corrected across seven tests.

Table 5. Proportion of electrodes showing changes in PAC due to theta waveform properties while viewing Mismatch objects.

	N	Power	Sin(θ)	Cos(θ)	S_Peak	S_Trough	Asym	Full
S1	8	0.25	0	0.13	0	0.13	0.38^*	0.13
S2	4	0	0.25	0.25	0.25	0	0	0.25
S3	6	0.33	0	0	0	0	0.17	0
S4	8	0.13	0	0	0	0.25	0.13	0
S5	6	0.33	0.17	0	0	0	0	0.17
Group	32	0.22^*	0.06	0.06	0.03	0.09	0.16	0.09

Open in a new tab

We next asked whether changes in these theta waveform properties contributed to the observed changes in PAC associated with associative novelty and memory updating. To confirm that our PAC analyses were not primarily caused by differences in theta amplitude (or any other oscillatory features such as peak/trough sharpness), we repeated our initial PAC analysis after regressing out variability explained by additional single-trial measures of theta amplitude, phase-locking, and waveform shape. The results of this analysis revealed the same relation between PAC and detection of associative novelty and memory updating. These control analyses rule out the possibility that observed differences in PAC resulted from changes to theta waveform shape, which is non-sinusoidal and known to vary with behavioral state (Cole and Voytek, 2017; Scheffer-Teixeira and Tort, 2016).

Discussion

We set out to examine the relationship between hippocampal theta oscillations and memory-driven viewing behaviors during visual exploration. To achieve this goal, we examined simultaneously recorded hippocampal potentials and eye movements while neurosurgical patients performed a spatial memory task. This paradigm allowed the disambiguation of eye movements driven by associative novelty (i.e. objects presented in updated spatial locations) and memory retrieval. We discovered that these two distinct viewing behaviors were uniquely tied to theta: phase-locking preceded retrieval-guided eye movements and followed novelty-driven fixations, predicting memory performance. These retrieval effects primarily occurred at the peak of theta, whereas novelty effects occurred at the trough of theta. Thus, we provide empirical support for models suggesting that interference between sensory information and memory representations is avoided by timing encoding and retrieval to occur at distinct phases of theta (Hasselmo et al., 2002). Analysis of theta to gamma PAC confirmed a potential mechanism by which the theta rhythm supports both encoding and retrieval. Modulation of gamma (80–100 Hz) amplitude by the theta rhythm increased during detection of updated object-locations. Further, increased theta to high-gamma PAC predicted memory updating, as determined by a subsequent memory test. These data support theories of active hippocampal involvement in visual exploration (Voss et al., 2017) and provide novel evidence that theta oscillations support both retrieval- and novelty-dependent viewing behaviors.

Based upon electrophysiological studies in humans and primates (Hoffman et al., 2013; Jutras et al., 2013), it has become apparent that hippocampal memory representations are used to guide saccades to behaviorally relevant locations (Meister and Buffalo, 2016). Of particular relevance to the present work, visual exploration of novel, but not repeated, scenes leads to a reset of hippocampal theta oscillations (Jutras et al., 2013). The consistency of this hippocampal phase reset predicts the success of novel encoding. In the present work, we provide evidence for conserved hippocampal processing in humans, by demonstrating increased phase consistency of theta oscillations following fixations to novel locations, which were further associated with improved memory performance. Recent work using fMRI (Liu et al., 2017) also suggests that visual exploration of scenes is associated with hippocampal function, as the number of fixations during encoding predicted both hippocampal activity and subsequent memory. This relationship between hippocampal activity and viewing behavior was limited to novel (but not repeated) stimuli, suggesting hippocampal activity and exploratory eye movements are linked specifically when encoding novel content. Using our spatial memory paradigm to identify retrieval-dependent eye movements, we found theta phase-locking occurred prior to saccade onset. Taken together with previous studies that have tied hippocampal activity to retrieval-guided eye movements (Bridge et al., 2017; Hannula and Ranganath, 2009) these results provide evidence that visual exploration is dependent upon the interplay of separate retrieval and novelty-detection mechanisms that underpin learning.

With the high temporal resolution of eye tracking and hippocampal potentials, our data are uniquely suited to clarify the hippocampal mechanisms that drive learning under associative novelty. Notably, we found increased theta phase-locking and power following fixations to mismatched spatial locations, consistent with previous hippocampal recordings in humans (Chen et al., 2013). The consistent timing of these effects following fixation onset provide further evidence that hippocampal theta is involved in the computation of mismatch signals, as opposed to providing signals of perceptual familiarity that could guide viewing to novel locations. This extends previous fMRI work which has established the involvement of the hippocampus in associative novelty detection (Kumaran and Maguire, 2006; Kumaran and Maguire, 2007a) and binding (Bridge and Voss, 2014b), specifically supporting the idea that the hippocampus acts as a comparator between new sensory inputs and prior memory (Kumaran and Maguire, 2007b; Lisman, 1999; Lisman and Grace, 2005; Vinogradova, 2001).

The observed interaction between the phase of hippocampal theta and amplitude of gamma oscillations provides insight into the mechanisms by which learning occurs under associative novelty, wherein sensory information conflicts with memory. Many theories propose that theta cycles segregate neuronal activity into functional packets based on coactivation of neuronal populations during distinct time windows (Buzsáki and Moser, 2013; Hasselmo et al., 2002; Rennó-Costa and Tort, 2017). These models allow for interleaved encoding and retrieval operations to occur at distinct phases of theta (Colgin, 2015), supported by nested higher frequency activity. In support of these models, we found that phase-locking associated with retrieval and novelty initiated at the peak and trough of hippocampal theta, respectively. In addition, theta to high-gamma PAC increased during the detection of novel spatial information, consistent with theta oscillations organizing high-frequency computations across hippocampal networks. In doing so, we show these models explain when memory-driven eye movements occur in humans, particularly when there are demands for memory encoding and retrieval to co-occur.

We found theta to gamma PAC predicted whether previously encoded memories would be updated to reflect current sensory information, in opposition to task demands. Prior work has shown that increased theta to gamma PAC in hippocampus reflects successful encoding (Lega et al., 2016). These findings may explain the cause of memory updating in our task, which was associated with forgetting of the original object-location. If hippocampal PAC supports encoding, increased PAC could strengthen associations between objects and updated locations. During the final memory test, the strength of these associations interferes with memory for the original location, resulting in forgetting. Despite this account, we cannot rule out the possibility that multiple processes contribute to memory updating. For example, updating could result from weakening of older memories during new learning or biased competition between memories at retrieval (Kuhl et al., 2012). Although our experimental design cannot distinguish between these accounts (i.e. memory updating necessarily causes forgetting of the original object-location), our findings clearly demonstrate that increased PAC predicts the formation of novel memories at the expense of prior learning.

Evaluation of theta waveform properties revealed that memory-related differences in PAC resulted from nested oscillations, as opposed to amplitude modulation of sharp waveforms or other changes to the theta waveform. Distinct neurophysiological states, defined by theta phase and modulation of gamma amplitude, indicated when viewing was driven by memory as opposed to novel information in the environment. While we did not observe theta-dependent modulation of gamma amplitude during retrieval, this could result from a number of methodological constrains, such as biased sampling to the anterior hippocampus and potential obscuring of narrowband gamma signals commonly observed in microelectrodes due to summation across larger neuronal populations.

Despite the emerging consensus in the rodent that theta oscillations in the hippocampus are responsible for segregating neuronal computations involved in encoding and retrieval operations (Colgin, 2015; Colgin, 2016; Hasselmo et al., 2002), translating these findings to observations in humans has proven challenging. Comparative studies between rodents and humans suggest that theta rhythms in the human are commonly observed at lower frequencies (e.g. 3 vs. 8 Hz) during exploratory behaviors (Watrous et al., 2013a). Additionally, studies have demonstrated multiple sources of theta rhythms in the medial temporal lobe (Lega et al., 2012; Mormann et al., 2008), including a low-theta or delta band (1–4 Hz) in addition to the typical theta band (4–8 Hz). Based on observations that encoding-related increases in power (Burke et al., 2013; Lega et al., 2012; Miller et al., 2018) and PAC (Lega et al., 2016) are specific to the low-theta band, it has been proposed that lower frequency theta in humans reflects a homologue of rodent theta (Jacobs, 2014). In contrast to this body of work, we observed faster (i.e. predominantly greater than 4 Hz) memory-related theta dynamics. We believe this difference stems from the emphasis on visual information in our task, resulting in a task-dependence in the frequency of theta oscillations. Higher frequency theta effects have been observed during visual search (Hoffman et al., 2013), in which alignment between saccades and theta oscillations was focused between 6–8 Hz. Our findings that fixations to objects caused theta phase resets in the 6–12 Hz range are consistent with work in nonhuman primates (Jutras et al., 2013), which demonstrated that saccades during visual exploration caused resets in hippocampal oscillations predominantly in the 8–11 Hz range. As in the present study, the magnitude of these resets predicted the success of encoding during memory formation. Multiple factors could account for changes in the speed of hippocampal theta, including the type of representation being processed in the medial temporal lobe (Watrous et al., 2013b), spatial attention required by a given task, or the rate of fixations. Future studies are necessary to determine what factors determine the speed of hippocampal theta oscillations and their relevance to different forms of memory.

Our findings of theta-dependent eye movements in the hippocampus are relevant to models of functional organization across the hippocampus. Drawn primarily from animal models, theories of hippocampal organization emphasize functional segregation along its long axis (Fanselow and Dong, 2010; Strange et al., 2014), with general agreement regarding a role of the anterior (ventral in rodents) hippocampus in emotion and affect, and the posterior (dorsal in rodents) hippocampus in spatial navigation and memory. As in most iEEG studies of hippocampal function in humans, we recorded primarily from the head and body of the hippocampus (see Figure 1b). Given the demands for spatial and mnemonic processing during our task, our results are not easily accommodated by these models. Our findings are consistent with recent theories that emphasize gradients in the scale of representations along the hippocampal long axis (Bellmund et al., 2018; Poppenk et al., 2013) with transitions from general to precise representations in the anterior to posterior direction. Thus, the emphasis of global spatial relations in our task (i.e. the relative positions of the original and updated locations rather than precise locations) may account for the observed effects within anterior hippocampus. Because our coverage was limited to the anterior aspect of the hippocampus, our study cannot directly address the anatomical specificity of memory-guided eye movements. However, modulation of hippocampal theta by eye movements has also been observed in the anterior hippocampus of non-human primates (Jutras et al., 2013), suggesting a conserved mechanism across primates.

While our analysis of theta oscillations was restricted to electrodes in the hippocampus (or adjacent white matter), memory-guided exploratory behaviors depend upon interactions between distributed cortical systems (Voss et al., 2017), particularly those involved in representing features within a scene, the spatial relations between these features, and transforming these memory representations into an oculomotor plan based on current visual input. Cells within the entorhinal cortex of macaques code the location of fixations in a grid-like fashion during free viewing (Killian et al., 2012), serving as a potential mechanism to provide a scale-invariant representation of fixation locations within the scene (Bicanski and Burgess, 2019). Similar grid-like modulation of entorhinal activity has been observed in humans using fMRI (Julian et al., 2018; Nau et al., 2018), providing converging evidence across species that the entorhinal system may provide a spatial framework for memory-guided viewing. As such, synchronous theta oscillations between the hippocampus and entorhinal systems would provide the spatial coding necessary to inform the oculomotor system of memory-relevant information. Although we were limited by electrode coverage, examining entorhinal-hippocampal synchrony and interactions between the hippocampus and other cortical systems should be a key aim for future studies.

One potential caveat is that the observed eye movements in this study do not reflect natural exploratory behaviors per-se but are rather driven by demands to learn and maintain the original object-location throughout the task. Thus, it is unclear how stereotyped these retrieval-dependent eye movements would be during unconstrained visual exploration. Free-viewing paradigms could build on this theoretical framework and determine the extent to which hippocampal-dependent viewing behaviors occur without task constraints. Given the observed retrieval phase-locking effects occurred well before saccade initiation, it is likely that the hippocampus plays a causal role in generating these eye movements. Causal manipulation of hippocampal theta, including stimulation-based approaches could be used to test this hypothesis, which is supported by growing evidence for disruptions in viewing behaviors from amnesic patients with hippocampal damage (Hannula et al., 2007; Lucas et al., 2019; Olsen et al., 2016; Ryan et al., 2000; Smith et al., 2006).

In conclusion, encoding and retrieval dependent eye movements are time locked to the phase of the hippocampal theta rhythm. Our findings support models wherein distinct phases of the theta cycle segregate neural processing of information related to encoding and retrieval (Colgin, 2016; Hasselmo et al., 2002; Hasselmo and Eichenbaum, 2005). Akin to spatial attention shifting between multiple locations relevant to a task at hand (Landau et al., 2015; Re et al., 2019), hippocampal theta could coordinate visual sampling between novel content in the environment and memory-rich spatial locations. The hippocampus thereby contributes to memory-guided behaviors through coordinated sampling of current and past perceptual states.

Materials and methods

Participants

Five subjects (three male; see Table 6 for demographic information) with refractory epilepsy performed our associative memory task during their stay at Northwestern Memorial Hospital (Chicago, IL). All subjects had depth electrodes implanted in the hippocampus as part of neurosurgical monitoring prior to elective surgery. Written informed consent was acquired from all subjects prior to participation in the research protocol in accordance with the Northwestern University Institutional Review Board.

Table 6. Subject demographics.

	S1	S2	S3	S4	S5
Age (years)	20	34	53	25	44
Sex	M	M	M	F	F
Full-scale IQ	94	109	105	121	92
Implanted Hemisphere	Left	Left	Right	Left	Left
Epileptic Focus	Basal temporal	Basal temporal	Middle hippocampus	Basal temporal	Amygdala
Etiology	Cortical dysplasia	Dysembryoplastic neuroepithelial tumor	Focal cortical dysplasia	Low grade glioma	Mesial temporal sclerosis
Duration of epilepsy (years)	10	10	8	3	41
Hippocampal contacts (n)	8	4	6	7	6

Open in a new tab

Experimental paradigm

We tested memory for associations between objects and their spatial locations using a novel spatial memory task. This task consisted of three distinct phases (Study, Refresh, Recognition), with each phase separated by a 60 s distractor (free viewing of scenes with domestic felines; Zhang et al., 2008). Subjects performed eight blocks in which they learned spatial locations for a sequence of 16 unique objects. Eye movements were recorded during each phase of the task, with five-point gaze calibration performed before each phase. Objects were 128 trial-unique images of real-life objects from the Bank of Standardized Stimuli (Brodeur et al., 2010). During each phase of the task, objects were presented at 3° of visual angle, with a red square of 0.2° of visual angle centered on each object. Stimuli were presented on a 23.6" monitor with a 120 Hz refresh rate from a stimulus control laptop. Synchronization pulses were sent from the stimulus control laptop to the clinical recording system using a DAQ control board, allowing alignment of electrophysiological and behavioral data.

At the beginning of each Study phase, a unique background image appeared for 5 s. Scenes provided visual information to assist learning unique spatial locations of 16 objects presented in the following block. Throughout the remainder of the Study phase, a sequence of 16 objects were presented at distinct locations superimposed on the background scene. At the start of each study trial, a fixation cross flashed twice on the screen (250 ms per flash, separated by 250 ms of the background scene) at the location of the next object. The fixation cross remained on the screen for a duration of 2 s, followed by presentation of the object for 3 s.

Next, subjects were tested on their spatial memory for each of the objects during the Refresh phase. During this phase of the task, three location cues indicated by small red squares (0.2° of visual angle) were presented in an equilateral triangle (randomly selected distance for each stimulus, mean distance of 12° and a range of 5.9–21.1° of visual angle across presented arrays). The object was presented at one of these three locations. Importantly, one of these locations was the object’s original location. On each block, half of the trials were randomly assigned to the Mismatch condition, in which the object was presented at one of the two novel locations. On the Match trials, the object was presented in its original location. Each trial began with the presentation of the background scene for 1 s followed by a fixation cross at the center of the screen for 1 s, at which point the object and location cues appeared for 5 s. Following stimulus presentation, memory for the original location of each item was tested. Subjects determined whether the object was in its original location, a new location, or if they were unsure by clicking a box that said: Same, Different, or Unsure. No feedback was given regarding the accuracy of each response.

Each block concluded with the Recognition phase which served as a final memory test for the original object-locations. The background scene appeared for 1 s, followed by the presentation of a fixation cross in the center of the screen for 1 s. Then, each object was presented at all three locations for a duration of 5 s. Following stimulus presentation, subjects selected the original object-location using a three-button response pad.

Eye tracking

Eye movements were recorded at 500 Hz using an Eyelink 1000 remote tracking system (SR Research, Ontario, Canada). Continuous eye-movement records were parsed into fixation, saccade, and blink events. Motion (0.15°), velocity (30°/s) and acceleration (8000°/s²) thresholds were used to identify saccade events. Blinks were identified based on pupil size, and remaining epochs below detection thresholds were classified as fixations. The location of each fixation event was computed as the average gaze position throughout the duration of the fixation. Circular viewing regions of interest (ROIs) were constructed based on a distance of 6° from one of the three potential object-locations. We focused our analysis of hippocampal activity to the subset of fixation events greater than 80 ms in duration. Except for our analyses of associative novelty (fixations to updated object-locations vs. repeated object-locations), which focused on early fixations to objects presented either in the original or updated object-location, we restricted our analysis to fixations that occurred 500 ms after object presentation and 500 ms before the end of each trial to avoid stimulus onset and offset effects.

Intracranial recordings

A combination of depth electrodes (Integra Life Sciences, Plainsboro NJ; AD-TECH Medical Instrument Co., Racine, WI; DIXI Medical, Besançon, France) as well as subdural grids and strips were implanted for neurosurgical monitoring. Our analyses focused on hippocampal depths, which had electrodes spaced 5 mm apart. Electrophysiological data were recorded to a clinical reference using a Nihon Kohden amplifier with a sample rate of 1–2 kHz with a bandpass filter from 0.6 to 600 Hz. Data were re-referenced to a bipolar montage and downsampled to 500 Hz as part of preprocessing. We analyzed bipolar pairs with at least one contact in hippocampal grey matter or proximal white matter. Line noise was reduced by application of a band-stop 4th order Butterworth filter. To rule out the possibility that epileptiform activity influenced our analyses, electrodes that exhibited inter-ictal spiking were excluded from analysis. In addition, all analyses were repeated after excluding contacts within the seizure onset zone (two electrodes in S3). The observed results were qualitatively identical, with no statistical differences when including all electrodes (all p > 0.05).

Anatomical localization

Post-implant CT (n = 4) or T1 weighted structural images (n = 1) were coregistered with presurgical T1 weighted structural MRIs using SPM12. Subdural electrodes were localized by reconstructing whole-brain cortical surfaces from pre-implant T1-weighted MRIs using the computational anatomy toolbox (Dahnke et al., 2013) and snapping electrode centroids to the cortical surface based on energy minimization (Dykstra et al., 2012). All T1-weighted MRI scans were normalized to MNI space by using a combination of affine and nonlinear registration steps, bias correction, and segmentation into grey matter, white matter, and cerebrospinal fluid components. Deformations from the normalization procedure were applied to individual electrode locations identified on post-implant CT images or structural images using Bioimage Suite (https://medicine.yale.edu/bioimaging/suite/).

Spectral decomposition

To examine oscillatory processes in the hippocampus, we decomposed bipolar recordings into measures of spectral phase and power using the continuous Morlet wavelet transform (wave number 5) across 30 logarithmically spaced frequencies from 1 to 10 Hz. We examined 1500 ms windows surrounding each fixation event of interest, with a 1250 ms buffer to prevent edge artifacts. Additional analysis of power spectra used the multitaper method to estimate spectral densities, ranging from 1 to 250 Hz. To identify oscillations, we modeled the power spectra as a mixture of two components: an aperiodic component modeled with an exponential, and putative oscillatory components modeled with Gaussians (Haller et al., 2018). The presence of oscillatory components was then evaluated by testing the equality of variances between aperiodic and full models.

Phase-locking analyses

We examined the relationship between the hippocampal theta rhythm and individual eye movements by computing the inter-trial phase coherence (a measure of phase-locking):

I T C_{f t} = \frac{1}{N} | \sum_{k = 1}^{N} e^{i φ f t k} |

for a given time (t) and frequency (f), where N is the total number of individual trials, k, and e^iφ is the polar representation of the phase angle, φ. This measure was computed separately for individual conditions of interest (e.g. fixations to a specific region of interest on the display). As this measure is biased by the number of observations, with fewer observations leading to inflated ITC measures, we used a random subsampling approach to ensure that the number of observations were matched prior to statistical testing. In addition to examining differences in the magnitude of phase-locking, we compared differences preferred phase angles, across conditions of interest. We estimated a difference in phase angles using the Watson-Williams test (Berens, 2009), followed by nonparametric statistical testing to assess significance.

Phase amplitude coupling analyses

Cross-frequency coupling between the phase of theta and gamma amplitude was computed using MI (Tort et al., 2010). MI is defined as the deviation in an amplitude distribution (across phases) from a uniform distribution, an adaptation from Kullback-Leibler distance (Kullback and Leibler, 1951), D_KL, that normalizes the range of the distance between zero and one:

M I = \frac{D_{K L} (P, U)}{l o g (N)}

where P is the normalized amplitude distribution as a function of phase, U is a uniform distribution, and N is the number of phase bins. For all presented analyses, we used 20 phase bins of 18°. MI takes values greater than zero when the observed amplitude varies with phase and is equal to zero when the distribution is uniform.

To circumvent the relatively short epochs in which we analyzed cross-frequency coupling (constrained by the frequency of eye movements during our task), we computed a standardized measure of the modulation index, MI_Z, via a surrogate control analysis. Specifically, for each trial and frequency combination, we permuted the observed phase timeseries across trials (separately for each condition of interest). This procedure was repeated 1000 times, resulting in a null distribution of MI values that could be explained by random (or condition-evoked) variations in the observed signal rather than true coupling between theta phase and gamma amplitude. MI_Z was measured as the difference between the observed MI and mean of the surrogate distribution, in units of standard deviations. These measures were used for all subsequent analysis of cross-frequency coupling.

We performed a series of follow-up analyses to rule out the possibility that memory-related changes in PAC were driven by changes in oscillatory power, phase-locking, or changes in the shape of theta oscillations. For each fixation of interest, we filtered (4th order Butterworth filter) bipolar signal into theta (1 to 10 Hz). Peak and trough amplitudes were extracted from the theta signal. Peak and trough sharpness were extracted as the average change in amplitude 2 ms before and after the inflection point. Theta amplitude and phase were extracted from the Hilbert transform of the filtered signal. As a single-trial estimate of phase-locking, we computed the difference in theta phase for each event from the average phase for each condition of interest. Trial-level measures were computed by averaging across all peaks or troughs within a given time window.

We used multiple regression to identify potential linear relations between properties of theta waveforms and trial-level measures of PAC (i.e. MI_Z). Phase angles were transformed into linear components via sine and cosine transform prior to regression. In addition to assessing the significance of these relations, we computed an adjusted measure of MI_Z that was independent of these theta waveform properties (the residuals of the linear model). Statistical evaluation of PAC was repeated after removing variance related to theta waveform properties.

Statistical analyses

We adopted a nonparametric permutation-based approach (Maris and Oostenveld, 2007) to correct for multiple comparisons across time and frequencies. When comparing differences in ITC or power between different types of fixations, we constructed a null distribution of differences by permuting the assignment of condition labels, blocked at the subject level. This null distribution was used to define an independent cluster-forming threshold for each observed measure (e.g. ITC at a specific time-frequency pair). When measures would be biased by the number of observations per condition (e.g. differences in ITC), random subsampling was used to equate the number of observations per condition. Individual clusters were considered significant (P_FWE < 0.05) if the summed statistic within each observed cluster exceeded 95% or 97.5% of clusters in the null distribution for one- and two-tailed tests, respectively. For tests comparing the relationship between the phase of an oscillation and spectral power, null distributions were constructed by permuting the phase timeseries across trials within each condition, per electrode and subject. These null distributions were used to standardize measures of phase amplitude coupling prior to statistical testing, as described above.

This permutation procedure also determined statistical significance of oscillations present at individual electrodes. After computing an F-statistic measuring the variance in aperiodic and full power spectra (see Spectral decomposition), we generated permutation distributions (n = 10,000) by randomizing condition labels (i.e. raw or aperiodic signals) across fixations. Statistical tests at the subject and group level were performed using binomial tests, comparing the proportion of significant electrodes to chance (determined to be 0.05 by the null distribution). We set a threshold of p < 0.05 for statistical significance, using Bonferroni correction to control for multiple comparisons across conditions and time periods.

Our sample size was determined based on prior work in humans and nonhuman primates (Hoffman et al., 2013; Jutras et al., 2013; Staudigl et al., 2017), which reported robust modulations in hippocampal theta due to visual sampling with a sample sizes ranging from two to six subjects. To support future meta-analytic work, we report Hedges’ g for dependent samples (Hentschke and Stüttgen, 2011) and proportion differences as measures of effect size where relevant. For cluster-based statistics, we estimated effect size by computing average effects within significant clusters on a per-subject basis (i.e. averaging across electrodes, frequency, and/or time) before group-level analysis.

Acknowledgements

We are grateful to Dr. Christina Zelano for helpful discussions, Irena Bellinski for assistance with patient recruitment, and the Laboratory of Human Neuroscience at Northwestern University for sharing resources; DJB acknowledges the support of NIH/National Institute of Mental Health (NIMH; grant R21MH115366) and National Center for Advancing Translational Sciences, Grant Number UL1TR001422. JEK was supported in part by National Institute of Neurological Disorders and Stroke grant T32NS047987.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

James E Kragel, Email: james.kragel@northwestern.edu.

Donna J Bridge, Email: donnajb@gmail.com.

Laura L Colgin, University of Texas at Austin, United States.

Funding Information

This paper was supported by the following grants:

National Institute of Mental Health R21MH115366 to Donna J Bridge.
National Center for Advancing Translational Sciences UL1TR001422 to Donna J Bridge.
National Institute of Neurological Disorders and Stroke T32NS047987 to James E Kragel.

Additional information

Competing interests

No competing interests declared.

Author contributions

Software, Formal analysis, Investigation, Visualization.

Resources.

Investigation.

Conceptualization, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology.

Ethics

Human subjects: Both verbal and written informed consent was obtained from all subjects prior to participation. This work was done in accordance with Northwestern University Institutional Review Board (IRB #: STU00202828).

Additional files

Supplementary file 1. Number of fixations for each comparison of interest.

elife-52108-supp1.docx^{(16.9KB, docx)}

Transparent reporting form

elife-52108-transrepform.docx^{(245.9KB, docx)}

Data availability

Behavioral data, eye movement data, continuous EEG recordings, and electrode locations in MNI space have been uploaded to NDAR.

The following dataset was generated:

Kragel JE, VanHaerents S, Templer JW, Schuele S, Rosenow JM, Nilakantan AS, Bridge DJ. 2019. Simultaneous eye tracking and hippocampal iEEG to identify oscillatory signals of memory retrieval and novelty detection in humans. NIMH Data Archive. 2890

References

Aru J, Aru J, Priesemann V, Wibral M, Lana L, Pipa G, Singer W, Vicente R. Untangling cross-frequency coupling in neuroscience. Current Opinion in Neurobiology. 2015;31:51–61. doi: 10.1016/j.conb.2014.08.002. [DOI] [PubMed] [Google Scholar]
Axmacher N, Elger CE, Fell J. Ripples in the medial temporal lobe are relevant for human memory consolidation. Brain. 2008;131:1806–1817. doi: 10.1093/brain/awn103. [DOI] [PubMed] [Google Scholar]
Axmacher N, Henseler MM, Jensen O, Weinreich I, Elger CE, Fell J. Cross-frequency coupling supports multi-item working memory in the human hippocampus. PNAS. 2010;107:3228–3233. doi: 10.1073/pnas.0911531107. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bellmund JLS, Gärdenfors P, Moser EI, Doeller CF. Navigating cognition: spatial codes for human thinking. Science. 2018;362:eaat6766. doi: 10.1126/science.aat6766. [DOI] [PubMed] [Google Scholar]
Berens P. CircStat: a MATLAB toolbox for circular statistics. Journal of Statistical Software. 2009;31:1–21. doi: 10.18637/jss.v031.i10. [DOI] [Google Scholar]
Bicanski A, Burgess N. A computational model of visual recognition memory via grid cells. Current Biology. 2019;29:979–990. doi: 10.1016/j.cub.2019.01.077. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bragin A, Jandó G, Nádasdy Z, Hetke J, Wise K, Buzsáki G. Gamma (40-100 hz) oscillation in the hippocampus of the behaving rat. The Journal of Neuroscience. 1995;15:47–60. doi: 10.1523/JNEUROSCI.15-01-00047.1995. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bridge DJ, Cohen NJ, Voss JL. Distinct hippocampal versus frontoparietal network contributions to retrieval and Memory-guided Exploration. Journal of Cognitive Neuroscience. 2017;29:1324–1338. doi: 10.1162/jocn_a_01143. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bridge DJ, Paller KA. Neural correlates of reactivation and retrieval-induced distortion. Journal of Neuroscience. 2012;32:12144–12151. doi: 10.1523/JNEUROSCI.1378-12.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bridge DJ, Voss JL. Active retrieval facilitates across-episode binding by modulating the content of memory. Neuropsychologia. 2014a;63:154–164. doi: 10.1016/j.neuropsychologia.2014.08.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bridge DJ, Voss JL. Hippocampal binding of novel information with dominant memory traces can support both memory stability and change. Journal of Neuroscience. 2014b;34:2203–2213. doi: 10.1523/JNEUROSCI.3819-13.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bridge DJ, Voss JL. Binding among select episodic elements is altered via active short-term retrieval. Learning & Memory. 2015;22:360–363. doi: 10.1101/lm.038703.115. [DOI] [PMC free article] [PubMed] [Google Scholar]
Brodeur MB, Dionne-Dostie E, Montreuil T, Lepage M. The bank of standardized stimuli (BOSS), a new set of 480 normative photos of objects to be used as visual stimuli in cognitive research. PLOS ONE. 2010;5:e10773. doi: 10.1371/journal.pone.0010773. [DOI] [PMC free article] [PubMed] [Google Scholar]
Burke JF, Zaghloul KA, Jacobs J, Williams RB, Sperling MR, Sharan AD, Kahana MJ. Synchronous and asynchronous theta and gamma activity during episodic memory formation. Journal of Neuroscience. 2013;33:292–304. doi: 10.1523/JNEUROSCI.2057-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
Buzsáki G, Moser EI. Memory, navigation and theta rhythm in the hippocampal-entorhinal system. Nature Neuroscience. 2013;16:130–138. doi: 10.1038/nn.3304. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chen J, Dastjerdi M, Foster BL, LaRocque KF, Rauschecker AM, Parvizi J, Wagner AD. Human hippocampal increases in low-frequency power during associative prediction violations. Neuropsychologia. 2013;51:2344–2351. doi: 10.1016/j.neuropsychologia.2013.03.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cole SR, van der Meij R, Peterson EJ, de Hemptinne C, Starr PA, Voytek B. Nonsinusoidal beta oscillations reflect cortical pathophysiology in Parkinson's Disease. The Journal of Neuroscience. 2017;37:4830–4840. doi: 10.1523/JNEUROSCI.2208-16.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cole SR, Voytek B. Brain oscillations and the importance of waveform shape. Trends in Cognitive Sciences. 2017;21:137–149. doi: 10.1016/j.tics.2016.12.008. [DOI] [PubMed] [Google Scholar]
Colgin LL. Do slow and fast gamma rhythms correspond to distinct functional states in the hippocampal network? Brain Research. 2015;1621:309–315. doi: 10.1016/j.brainres.2015.01.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
Colgin LL. Rhythms of the hippocampal network. Nature Reviews Neuroscience. 2016;17:239–249. doi: 10.1038/nrn.2016.21. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dahnke R, Yotter RA, Gaser C. Cortical thickness and central surface estimation. NeuroImage. 2013;65:336–348. doi: 10.1016/j.neuroimage.2012.09.050. [DOI] [PubMed] [Google Scholar]
Douchamps V, Jeewajee A, Blundell P, Burgess N, Lever C. Evidence for encoding versus retrieval scheduling in the hippocampus by theta phase and acetylcholine. Journal of Neuroscience. 2013;33:8689–8704. doi: 10.1523/JNEUROSCI.4483-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
Duncan K, Curtis C, Davachi L. Distinct memory signatures in the hippocampus: intentional states distinguish match and mismatch enhancement signals. Journal of Neuroscience. 2009;29:131–139. doi: 10.1523/JNEUROSCI.2998-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
Duncan K, Ketz N, Inati SJ, Davachi L. Evidence for area CA1 as a match/mismatch detector: a high-resolution fMRI study of the human hippocampus. Hippocampus. 2012;22:389–398. doi: 10.1002/hipo.20933. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dykstra AR, Chan AM, Quinn BT, Zepeda R, Keller CJ, Cormier J, Madsen JR, Eskandar EN, Cash SS. Individualized localization and cortical surface-based registration of intracranial electrodes. NeuroImage. 2012;59:3563–3570. doi: 10.1016/j.neuroimage.2011.11.046. [DOI] [PMC free article] [PubMed] [Google Scholar]
Fanselow MS, Dong HW. Are the dorsal and ventral hippocampus functionally distinct structures? Neuron. 2010;65:7–19. doi: 10.1016/j.neuron.2009.11.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
Haller M, Donoghue T, Peterson E, Varma P, Sebastian P, Gao R, Noto T, Knight RT, Shestyuk A, Voytek B. Parameterizing neural power spectra. bioRxiv. 2018 doi: 10.1101/299859. [DOI] [PMC free article] [PubMed]
Hannula DE, Ryan JD, Tranel D, Cohen NJ. Rapid onset relational memory effects are evident in eye movement behavior, but not in hippocampal amnesia. Journal of Cognitive Neuroscience. 2007;19:1690–1705. doi: 10.1162/jocn.2007.19.10.1690. [DOI] [PubMed] [Google Scholar]
Hannula DE, Ranganath C. The eyes have it: hippocampal activity predicts expression of memory in eye movements. Neuron. 2009;63:592–599. doi: 10.1016/j.neuron.2009.08.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hasselmo ME, Bodelón C, Wyble BP. A proposed function for hippocampal theta rhythm: separate phases of encoding and retrieval enhance reversal of prior learning. Neural Computation. 2002;14:793–817. doi: 10.1162/089976602317318965. [DOI] [PubMed] [Google Scholar]
Hasselmo ME, Eichenbaum H. Hippocampal mechanisms for the context-dependent retrieval of episodes. Neural Networks. 2005;18:1172–1190. doi: 10.1016/j.neunet.2005.08.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hentschke H, Stüttgen MC. Computation of measures of effect size for neuroscience data sets. European Journal of Neuroscience. 2011;34:1887–1894. doi: 10.1111/j.1460-9568.2011.07902.x. [DOI] [PubMed] [Google Scholar]
Heusser AC, Poeppel D, Ezzyat Y, Davachi L. Episodic sequence memory is supported by a theta-gamma phase code. Nature Neuroscience. 2016;19:1374–1380. doi: 10.1038/nn.4374. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hoffman KL, Dragan MC, Leonard TK, Micheli C, Montefusco-Siegmund R, Valiante TA. Saccades during visual exploration align hippocampal 3-8 hz rhythms in human and non-human primates. Frontiers in Systems Neuroscience. 2013;7:43. doi: 10.3389/fnsys.2013.00043. [DOI] [PMC free article] [PubMed] [Google Scholar]
Honey RC, Watt A, Good M. Hippocampal lesions disrupt an associative mismatch process. The Journal of Neuroscience. 1998;18:2226–2230. doi: 10.1523/JNEUROSCI.18-06-02226.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
Howard LR, Kumaran D, Olafsdottir HF, Spiers HJ. Double dissociation between hippocampal and parahippocampal responses to object-background context and scene novelty. Journal of Neuroscience. 2011;31:5253–5261. doi: 10.1523/JNEUROSCI.6055-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
Jacobs J. Hippocampal theta oscillations are slower in humans than in rodents: implications for models of spatial navigation and memory. Philosophical Transactions of the Royal Society B: Biological Sciences. 2014;369:20130304. doi: 10.1098/rstb.2013.0304. [DOI] [PMC free article] [PubMed] [Google Scholar]
Julian JB, Keinath AT, Frazzetta G, Epstein RA. Human entorhinal cortex represents visual space using a boundary-anchored grid. Nature Neuroscience. 2018;21:191–194. doi: 10.1038/s41593-017-0049-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
Jutras MJ, Fries P, Buffalo EA. Oscillatory activity in the monkey hippocampus during visual exploration and memory formation. PNAS. 2013;110:13144–13149. doi: 10.1073/pnas.1302351110. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kahana MJ, Seelig D, Madsen JR. Theta returns. Current Opinion in Neurobiology. 2001;11:739–744. doi: 10.1016/S0959-4388(01)00278-1. [DOI] [PubMed] [Google Scholar]
Killian NJ, Jutras MJ, Buffalo EA. A map of visual space in the primate entorhinal cortex. Nature. 2012;491:761–764. doi: 10.1038/nature11587. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kuhl BA, Bainbridge WA, Chun MM. Neural reactivation reveals mechanisms for updating memory. Journal of Neuroscience. 2012;32:3453–3461. doi: 10.1523/JNEUROSCI.5846-11.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kullback S, Leibler RA. On information and sufficiency. The Annals of Mathematical Statistics. 1951;22:79–86. doi: 10.1214/aoms/1177729694. [DOI] [Google Scholar]
Kumaran D, Maguire EA. An unexpected sequence of events: mismatch detection in the human Hippocampus. PloS Biology. 2006;4:e424. doi: 10.1371/journal.pbio.0040424. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kumaran D, Maguire EA. Match mismatch processes underlie human hippocampal responses to associative novelty. Journal of Neuroscience. 2007a;27:8517–8524. doi: 10.1523/JNEUROSCI.1677-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kumaran D, Maguire EA. Which computational mechanisms operate in the Hippocampus during novelty detection? Hippocampus. 2007b;17:735–748. doi: 10.1002/hipo.20326. [DOI] [PubMed] [Google Scholar]
Kumaran D, Maguire EA. Novelty signals: a window into hippocampal information processing. Trends in Cognitive Sciences. 2009;13:47–54. doi: 10.1016/j.tics.2008.11.004. [DOI] [PubMed] [Google Scholar]
Landau AN, Schreyer HM, van Pelt S, Fries P. Distributed attention is implemented through theta-rhythmic gamma modulation. Current Biology. 2015;25:2332–2337. doi: 10.1016/j.cub.2015.07.048. [DOI] [PubMed] [Google Scholar]
Lega BC, Jacobs J, Kahana M. Human hippocampal theta oscillations and the formation of episodic memories. Hippocampus. 2012;22:748–761. doi: 10.1002/hipo.20937. [DOI] [PubMed] [Google Scholar]
Lega B, Burke J, Jacobs J, Kahana MJ. Slow-theta-to-gamma phase-amplitude coupling in human hippocampus supports the formation of new episodic memories. Cerebral Cortex. 2016;26:268–278. doi: 10.1093/cercor/bhu232. [DOI] [PMC free article] [PubMed] [Google Scholar]
Leonard TK, Mikkila JM, Eskandar EN, Gerrard JL, Kaping D, Patel SR, Womelsdorf T, Hoffman KL. Sharp wave ripples during visual exploration in the primate hippocampus. Journal of Neuroscience. 2015;35:14771–14782. doi: 10.1523/JNEUROSCI.0864-15.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
Leonard TK, Hoffman KL. Sharp-wave ripples in primates are enhanced near remembered visual objects. Current Biology. 2017;27:257–262. doi: 10.1016/j.cub.2016.11.027. [DOI] [PubMed] [Google Scholar]
Lever C, Burton S, Jeewajee A, Wills TJ, Cacucci F, Burgess N, O'Keefe J. Environmental novelty elicits a later theta phase of firing in CA1 but not subiculum. Hippocampus. 2010;20:229–234. doi: 10.1002/hipo.20671. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lisman JE. Relating hippocampal circuitry to function: recall of memory sequences by reciprocal dentate-CA3 interactions. Neuron. 1999;22:233–242. doi: 10.1016/s0896-6273(00)81085-5. [DOI] [PubMed] [Google Scholar]
Lisman JE, Grace AA. The hippocampal-VTA loop: controlling the entry of information into long-term memory. Neuron. 2005;46:703–713. doi: 10.1016/j.neuron.2005.05.002. [DOI] [PubMed] [Google Scholar]
Lisman JE, Jensen O. The theta-gamma neural code. Neuron. 2013;77:1002–1016. doi: 10.1016/j.neuron.2013.03.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
Liu ZX, Shen K, Olsen RK, Ryan JD. Visual sampling predicts hippocampal activity. The Journal of Neuroscience. 2017;37:599–609. doi: 10.1523/JNEUROSCI.2610-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Long NM, Lee H, Kuhl BA. Hippocampal mismatch signals are modulated by the strength of neural predictions and their similarity to outcomes. The Journal of Neuroscience. 2016;36:12677–12687. doi: 10.1523/JNEUROSCI.1850-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lopes-Dos-Santos V, van de Ven GM, Morley A, Trouche S, Campo-Urriza N, Dupret D. Parsing hippocampal theta oscillations by nested spectral components during spatial exploration and memory-guided behavior. Neuron. 2018;100:940–952. doi: 10.1016/j.neuron.2018.09.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lucas HD, Duff MC, Cohen NJ. The hippocampus promotes effective saccadic information gathering in humans. Journal of Cognitive Neuroscience. 2019;31:186–201. doi: 10.1162/jocn_a_01336. [DOI] [PubMed] [Google Scholar]
Manns JR, Zilli EA, Ong KC, Hasselmo ME, Eichenbaum H. Hippocampal CA1 spiking during encoding and retrieval: relation to theta phase. Neurobiology of Learning and Memory. 2007;87:9–20. doi: 10.1016/j.nlm.2006.05.007. [DOI] [PubMed] [Google Scholar]
Maris E, Oostenveld R. Nonparametric statistical testing of EEG- and MEG-data. Journal of Neuroscience Methods. 2007;164:177–190. doi: 10.1016/j.jneumeth.2007.03.024. [DOI] [PubMed] [Google Scholar]
Meister MLR, Buffalo EA. Getting directions from the hippocampus: the neural connection between looking and memory. Neurobiology of Learning and Memory. 2016;134:135–144. doi: 10.1016/j.nlm.2015.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
Miller J, Watrous AJ, Tsitsiklis M, Lee SA, Sheth SA, Schevon CA, Smith EH, Sperling MR, Sharan A, Asadi-Pooya AA, Worrell GA, Meisenhelter S, Inman CS, Davis KA, Lega B, Wanda PA, Das SR, Stein JM, Gorniak R, Jacobs J. Lateralized hippocampal oscillations underlie distinct aspects of human spatial memory and navigation. Nature Communications. 2018;9:2423. doi: 10.1038/s41467-018-04847-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mormann F, Fell J, Axmacher N, Weber B, Lehnertz K, Elger CE, Fernández G. Phase/amplitude reset and theta-gamma interaction in the human medial temporal lobe during a continuous word recognition memory task. Hippocampus. 2005;15:890–900. doi: 10.1002/hipo.20117. [DOI] [PubMed] [Google Scholar]
Mormann F, Osterhage H, Andrzejak RG, Weber B, Fernández G, Fell J, Elger CE, Lehnertz K. Independent Delta/theta rhythms in the human hippocampus and entorhinal cortex. Frontiers in Human Neuroscience. 2008;2:3. doi: 10.3389/neuro.09.003.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
Morris RG, Inglis J, Ainge JA, Olverman HJ, Tulloch J, Dudai Y, Kelly PA. Memory reconsolidation: sensitivity of spatial memory to inhibition of protein synthesis in dorsal hippocampus during encoding and retrieval. Neuron. 2006;50:479–489. doi: 10.1016/j.neuron.2006.04.012. [DOI] [PubMed] [Google Scholar]
Nau M, Navarro Schröder T, Bellmund JLS, Doeller CF. Hexadirectional coding of visual space in human entorhinal cortex. Nature Neuroscience. 2018;21:188–190. doi: 10.1038/s41593-017-0050-8. [DOI] [PubMed] [Google Scholar]
Newman EL, Gillet SN, Climer JR, Hasselmo ME. Cholinergic blockade reduces theta-gamma phase amplitude coupling and speed modulation of theta frequency consistent with behavioral effects on encoding. Journal of Neuroscience. 2013;33:19635–19646. doi: 10.1523/JNEUROSCI.2586-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
Olsen RK, Sebanayagam V, Lee Y, Moscovitch M, Grady CL, Rosenbaum RS, Ryan JD. The relationship between eye movements and subsequent recognition: evidence from individual differences and amnesia. Cortex. 2016;85:182–193. doi: 10.1016/j.cortex.2016.10.007. [DOI] [PubMed] [Google Scholar]
Poppenk J, Evensmoen HR, Moscovitch M, Nadel L. Long-axis specialization of the human hippocampus. Trends in Cognitive Sciences. 2013;17:230–240. doi: 10.1016/j.tics.2013.03.005. [DOI] [PubMed] [Google Scholar]
Re D, Inbar M, Richter CG, Landau AN. Feature-based attention samples stimuli rhythmically. Current Biology. 2019;29:693–699. doi: 10.1016/j.cub.2019.01.010. [DOI] [PubMed] [Google Scholar]
Rennó-Costa C, Tort ABL. Place and grid cells in a loop: implications for memory function and spatial coding. The Journal of Neuroscience. 2017;37:8062–8076. doi: 10.1523/JNEUROSCI.3490-16.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ryan JD, Althoff RR, Whitlow S, Cohen NJ. Amnesia is a deficit in relational memory. Psychological Science. 2000;11:454–461. doi: 10.1111/1467-9280.00288. [DOI] [PubMed] [Google Scholar]
Scheffer-Teixeira R, Tort AB. On cross-frequency phase-phase coupling between theta and gamma oscillations in the hippocampus. eLife. 2016;5:e20515. doi: 10.7554/eLife.20515. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sederberg PB, Kahana MJ, Howard MW, Donner EJ, Madsen JR. Theta and gamma oscillations during encoding predict subsequent recall. The Journal of Neuroscience. 2003;23:10809–10814. doi: 10.1523/JNEUROSCI.23-34-10809.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
Shirhatti V, Borthakur A, Ray S. Effect of reference scheme on power and phase of the local field potential. Neural Computation. 2016;28:882–913. doi: 10.1162/NECO_a_00827. [DOI] [PMC free article] [PubMed] [Google Scholar]
Siegle JH, Wilson MA. Enhancement of encoding and retrieval functions through theta phase-specific manipulation of hippocampus. eLife. 2014;3:e03061. doi: 10.7554/eLife.03061. [DOI] [PMC free article] [PubMed] [Google Scholar]
Smith CN, Hopkins RO, Squire LR. Experience-dependent eye movements, awareness, and hippocampus-dependent memory. Journal of Neuroscience. 2006;26:11304–11312. doi: 10.1523/JNEUROSCI.3071-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
Staresina BP, Bergmann TO, Bonnefond M, van der Meij R, Jensen O, Deuker L, Elger CE, Axmacher N, Fell J. Hierarchical nesting of slow oscillations, spindles and ripples in the human hippocampus during sleep. Nature Neuroscience. 2015;18:1679–1686. doi: 10.1038/nn.4119. [DOI] [PMC free article] [PubMed] [Google Scholar]
Staudigl T, Hartl E, Noachtar S, Doeller CF, Jensen O. Saccades are phase-locked to alpha oscillations in the occipital and medial temporal lobe during successful memory encoding. PloS Biology. 2017;15:e2003404. doi: 10.1371/journal.pbio.2003404. [DOI] [PMC free article] [PubMed] [Google Scholar]
Strange BA, Witter MP, Lein ES, Moser EI. Functional organization of the hippocampal longitudinal axis. Nature Reviews Neuroscience. 2014;15:655–669. doi: 10.1038/nrn3785. [DOI] [PubMed] [Google Scholar]
Thakral PP, Yu SS, Rugg MD. The hippocampus is sensitive to the mismatch in novelty between items and their contexts. Brain Research. 2015;1602:144–152. doi: 10.1016/j.brainres.2015.01.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
Tort AB, Komorowski RW, Manns JR, Kopell NJ, Eichenbaum H. Theta-gamma coupling increases during the learning of item-context associations. PNAS. 2009;106:20942–20947. doi: 10.1073/pnas.0911331106. [DOI] [PMC free article] [PubMed] [Google Scholar]
Tort AB, Komorowski R, Eichenbaum H, Kopell N. Measuring phase-amplitude coupling between neuronal oscillations of different frequencies. Journal of Neurophysiology. 2010;104:1195–1210. doi: 10.1152/jn.00106.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
Vaz AP, Yaffe RB, Wittig JH, Inati SK, Zaghloul KA. Dual origins of measured phase-amplitude coupling reveal distinct neural mechanisms underlying episodic memory in the human cortex. NeuroImage. 2017;148:148–159. doi: 10.1016/j.neuroimage.2017.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
Vaz AP, Inati SK, Brunel N, Zaghloul KA. Coupled ripple oscillations between the medial temporal lobe and neocortex retrieve human memory. Science. 2019;363:975–978. doi: 10.1126/science.aau8956. [DOI] [PMC free article] [PubMed] [Google Scholar]
Vinogradova OS. Hippocampus as comparator: role of the two input and two output systems of the hippocampus in selection and registration of information. Hippocampus. 2001;11:578–598. doi: 10.1002/hipo.1073. [DOI] [PubMed] [Google Scholar]
Voss JL, Warren DE, Gonsalves BD, Federmeier KD, Tranel D, Cohen NJ. Spontaneous revisitation during visual exploration as a link among strategic behavior, learning, and the hippocampus. PNAS. 2011;108:E402–E409. doi: 10.1073/pnas.1100225108. [DOI] [PMC free article] [PubMed] [Google Scholar]
Voss JL, Bridge DJ, Cohen NJ, Walker JA. A closer look at the hippocampus and memory. Trends in Cognitive Sciences. 2017;21:577–588. doi: 10.1016/j.tics.2017.05.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
Watrous AJ, Lee DJ, Izadi A, Gurkoff GG, Shahlaie K, Ekstrom AD. A comparative study of human and rat hippocampal low-frequency oscillations during spatial navigation. Hippocampus. 2013a;23:656–661. doi: 10.1002/hipo.22124. [DOI] [PMC free article] [PubMed] [Google Scholar]
Watrous AJ, Tandon N, Conner CR, Pieters T, Ekstrom AD. Frequency-specific network connectivity increases underlie accurate spatiotemporal memory retrieval. Nature Neuroscience. 2013b;16:349–356. doi: 10.1038/nn.3315. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wilming N, Onat S, Ossandón JP, Açık A, Kietzmann TC, Kaspar K, Gameiro RR, Vormberg A, König P. An extensive dataset of eye movements during viewing of complex images. Scientific Data. 2017;4:160126. doi: 10.1038/sdata.2016.126. [DOI] [PMC free article] [PubMed] [Google Scholar]
Winters BD, Tucci MC, Jacklin DL, Reid JM, Newsome J. On the dynamic nature of the engram: evidence for circuit-level reorganization of object memory traces following reactivation. Journal of Neuroscience. 2011;31:17719–17728. doi: 10.1523/JNEUROSCI.2968-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zhang W, Sun J, Tang X. Cat head detection - How to effectively exploit shape and texture features. European Conference on Computer Vision; Marseille, France. 2008. pp. 802–816. [DOI] [Google Scholar]

eLife. doi: 10.7554/eLife.52108.sa1

Decision letter

Editor: Laura L Colgin¹

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Hippocampal theta coordinates memory processing during visual exploration" for consideration by eLife. Your article has been reviewed by a Senior Editor, a Reviewing Editor, and three reviewers. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

Reviewers all found the topic and experimental design compelling. However, all had significant concerns about the statistics and analyses that raised questions about whether the paper's conclusions were strongly supported by the results. The separate reviews are included below in their entirety, but major concerns that led to the rejection decision include:

1) The time-window for analysis: It is unclear that the data only include the fixation of interest (and not any other saccade or fixation). Changing this window would dramatically decrease the available data for analysis to ~250 ms for each fixation.

2) Improper statistics: Analyses lacked a direct test of the interaction of interest (and instead concluded differential effects based on the finding that one result was significant, and another was not).

3) The PAC results were viewed as a weak part of the paper. Concerns were raised that PAC effects may have resulted from changes in power or waveform asymmetry and that analyses were not corrected for multiple comparisons.

Differences in the gamma frequency observed (very fast) compared to previous results in rodents (e.g., Colgin et al.) made the discussion of the results as related to slow and fast gamma strange.

4) There was a lack of demonstration of peaks in the spectra for the frequency ranges that were chosen for analysis to justify why these particular frequencies were selected. It is possible that similar results observed at other frequencies and that drawing conclusions about specific oscillatory patterns are unwarranted.

5) Nothing is shown to describe inter-subject variability.

6) Pre- and post- fixation results were not shown for Match trials (i.e., similar to Figure 2).

If you decide to re-do your analyses and statistics, and the conclusions are supported by the new analyses, we would be willing to reconsider a resubmission of this manuscript.

Reviewer #1:Kragel et al. examine the relation between hippocampal field potentials and eye movements during a visuospatial memory task. The primary claims are that hippocampal theta oscillations at 5Hz exhibit phase-locking around the time of eye movements. Specifically, pre-saccade phase-locking increases for saccades to remembered locations whereas post-saccade phase-locking increases for new (updated) locations. They also reported that saccades to the locations of new stimuli were associated with increases in phase-amplitude coupling between theta and gamma activity.

I think the general topic of this paper is interesting but I have quite substantial concerns about whether the reported data back up the claims. Not all the findings are reported in sufficient detail and I see a number of statistical issues.

- The most substantial concern I see with the paper is that many of the paper's claims hinge on a comparison of two statistical tests, with the authors making much of the fact that one comparison is significant and that another is not. Instead the paper needs to directly test whether one effect is larger than the other, such as by testing the interaction with an ANOVA or some other way. This is a problem in many sections of the paper and substantially undermines the specificity of key claims. The authors should review all their claims and ensure that result is supported by a specific statistical test/interaction, rather than just relying on an effect being significant in one condition and not in another statistically.

- I was disappointed that the paper only measured the magnitude of phase-locking, while seemingly ignoring the key issue of what specific phase the locking occurred to. The models that they refer to from Hasselmo and Colgin have strong hypotheses related to specific theta phases and the recorded data would seem to measure this data. But unfortunately, the reported data analyses ignore testing which theta phase the locking occurs to, instead focusing only the magnitude of locking. Without this information, the paper doesn't really provide strong support for those theoretical models.

- It is notable that the latencies of the effects are close to 300ms post saccade. Are the effects related to P300s?

- The PAC analyses are not compelling because they do not rule out the possibility that the effects are driven by changes in power. Based on the data that they show, I think it remains possible that their PAC results are a direct result of changes in theta power or locking without any specific PAC changes. To substantiate their claim that their findings demonstrate a specific role for PAC in memory, the authors need to do much more to specifically show that their effects are caused by interactions between the timing of gamma and theta rather than power or phase changes in just one of the signals. They also should rule out whether the PAC changes could be related to waveform asymmetry.

- I had a hard time following the control analyses in the third paragraph of subsection “Theta dependence of memory-guided eye movements”. Can the text here be reworded so the logic is clearer?

- I was surprised to see high-gamma effects at frequencies as high as 130-150 Hz. Were these results corrected for multiple comparisons? In the Discussion section this pattern is compared to the fast and slow gamma oscillations seen in rodents by Colgin et al. However, this signal is so substantially faster in frequency than the signals seen in rodents that this comparison seems far-fetched.

- It seems that most of the paper's statistics are performed across subjects (based on the degrees of freedom). This is a useful, conservative approach, which they should better explain this in the results. They used a different approach in subsection “Theta to gamma phase amplitude coupling predicts memory updating”, when they report percentages of hippocampal electrodes that show each effect, which they should explain and justify.

Reviewer #2:

This manuscript uses an innovative analysis of memory-guided eye movements to examine the relationship between theta phase and encoding and retrieval processes through recordings from macro electrodes in the hippocampus in epilepsy patients. The authors report theta phase-locking prior to fixations to retrieved locations along with theta phase-locking following fixations to novel locations. Further, this phase-locking for novel (conflicting) information predicted memory for the original information. The authors also examined theta-gamma phase amplitude coupling (PAC) and identified increased PAC for fixations to updated locations compared to repeated locations, and that greater PAC was associated with worse subsequent memory performance, i.e., updating of the object location in memory. The manuscript addresses an important topic: the relationship between memory processing and theta phase, and uses an interesting and innovative behavioral measure. However, I have several concerns regarding the analyses and the presentation of the results that impact the clarity of the conclusions.

1) I have concerns about the time course of the behavior and the analysis windows chosen. Table 1 shows that fixations last ~200-400 ms, which is consistent with a broad literature. However, many of the analyses use a window which surely encompasses more than the fixation of interest. For example, Figure 2B, the significant cluster of ITC occurs begins ~400 ms prior to the fixation. It is likely that this time window includes the fixation 2-back from the fixation of interest. Similarly, the post-fixation effect shown in Figure 2C likely includes the fixation of interest plus the following fixation. It seems that the cleanest approach would be to limit the analysis to the immediately preceding or following fixation. This may have been the approach taken, but it is unclear from the methods. Similarly, it seems likely that multiple fixations within a given ROI occur in succession (as illustrated in Figure 1B). It is not clear how this would have been controlled for in the analysis.

2) Because the presentation of the data are for the most part, fairly processed, it would be helpful to show an example of the raw LFP and theta phase-locking in an individual fixation. It is also very difficult to understand how many fixations were included in each of the analyses.

3) For the PAC analysis, it is unclear how the example of the representative electrode fits with the population result. The representative electrode showed increased PAC between theta (~5 Hz) and gamma (~140-170 Hz) for fixations to updated vs repeated objects. By contrast, the population effects showed significant PAC at 80-100 Hz. From Figure 5A, it does look like there may be a small increase in PAC in the 140-170 range, which did not reach significance. Is it possible that different PAC effects were observed at different electrode locations?

4) In the Discussion section, the authors report that "inspection of raw traces revealed that the observed differences in PAC resulted from nested oscillations, as opposed to the modulation in the amplitude of sharp waveforms." This is an interesting finding, and it would be helpful to provide examples of these raw traces.

5) During the Refresh phase, the subjects performed a recognition memory task, but I couldn't find these results reported. Importantly, it was not clear whether the subjects received feedback on the accuracy of their responses and how their responses on the Refresh phase correlated with accuracy in the final Test phase.

6) In the Materials and methods section, it is not clear whether a distinct background was used for each of the 16 objects or if the same background was used throughout a block. If a different background was used, was each background shown for a 5s familiarization period prior to the Study phase? It was also not clear whether the flashing of the fixation cross was in the location of the object. Was there one study trial per object/scene? Did all subjects perform all 8 blocks?Reviewer #3:The study "Hippocampal theta coordinates memory processing during visual exploration" presents behavioral data showing that viewing probability on a novel stimulus location on mismatch trials predicts later recognition memory, while looking behavior on match trials did not. The authors then selected fixations to mismatch locations and found that they were preceded by theta phase consistency in hippocampal recordings (Figure 2). Compared to repeat fixations, the fixations to updated locations resulted in stronger post-fixational ~4.5 Hz ITS (Figure 3). Additionally the study reports enhanced 5 to ~90Hz phase amplitude coupling for updating fixations and decreased ~5-140Hz phase amplitude coupling to a new object when the original location was forgotten as if it indicated an overriding of memory for the old object location association.

This is an interesting study, reporting findings in the human medial temporal lobe that extend prior studies in human and nonhuman primates showing that more consistent fixation aligned phases in the theta band relate to the updating and better remembering of newly fixated objects.

The study uses a well balanced design and applies state-of-the-art methods.

Beyond these strengths there are several aspects of the analysis and writing that deserve consideration.

1) A major concern with this study is the restricted frequency range analyzed which limits interpretation and understanding of memory updating related dynamics. The reader is presented with phase analysis results of the 1-10 Hz range and the 80-200 Hz range. This restriction does not allow to discern how overall (average) and beta and gamma activity relate to memory updating despite prior studies implicating these frequency ranges in related functions. What is the average time-locked average ERP activity around fixations (does that allows to see a reset like behavior)? Are there peri-fixation effects of ITC effects at beta and gamma activities that are related to updating and remembering fixated objects?

2) Were there power spectral peaks discernible at those frequencies at which phases are interpreted? If not, can this be discussed explicitly? What were the shapes of the power spectra? If power peaks at non-existent or at frequencies away from the max-ITC phase effects that would be important to know to constrain interpretation and guide future studies aimed at finding the neural basis of the ITC effects.

3) Figure 2 (legend and main text) insinuate that the analyzed fixations are all "retrieval-dependent". I think this is a misleading statement because it is not made sufficiently clear which fixations were used and only a small fraction may relate to retrieval. Were only Mismatch fixation used? Only from trials which were later 'remembered'? Including the first fixation?

If the behavior suggest that there are ~2 sec where refixations to the mismatch locations are predictive to the remembering, then fixations at other time windows are not linked to later remembering. This should be ideally made clear to the reader.

4) Related to the previous point, it would be appreciated to also see and report the pre- and post-fixational ITC for Match trials (e.g. in a figure supplement similar to Figure 2). If you would have a similar 5Hz ITC in those conditions then the interpretation of the pre-fixational mismatch ITC would be different.

5) It is unclear which "additional" features are possibly explored (given similar fixation durations) that is suggested by this sentence (subsection “Theta dependence of memory-guided eye movements”): " the number of fixations to the object was reduced on Mismatch relative to Match trials (paired t-test, t4 = -8.8, P = 0.0009, g = -0.6), indicating that subjects explored additional visual features during Mismatch".

6) With five subjects it seems necessary to report about the variability and consistency of the observed main findings more explicitly. How many subjects showed a positive memory effect for mismatch fixation trials (at which specific anatomical locations)? In how many subjects was there a predictive theta ITC? In how many subjects was there a post-fixational memory ITC effect?

7) How often were individual objects of the 16 objects shown (eight times?)? Were they shown the same numbers and durations? If no, then please test if this affects the theta ITC effect

8) Were objects shown in the same sequence in each of the eight blocks? If yes, please show that sequence effects are not explaining the memory predictive theta ITC.

9) Were the 16 objects shown at 16 random locations or were some shown at the same location? If the latter is the case, please provide test that the number of objects per location is not a confound.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Hippocampal theta coordinates memory processing during visual exploration" for consideration by eLife. Your article has been reviewed by Laura Colgin as the Senior Editor, a Reviewing Editor, and three reviewers. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This paper reports that eye movements are phase-locked to hippocampal theta rhythms in humans performing a spatial memory task. A major strength of the paper is its close link to theoretical predictions of distinct hippocampal theta phases being linked to novelty-related exploration/encoding and retrieval. The results provide quantitative data supporting this model in human subjects. However, there are several issues that require clarification and revisions that should be made to increase the impact of this work on a broad range of readers. Several features of the analyses need to be clarified for the results to be considered solid. In general, the authors should attempt to write up the results more clearly in a way that can be easily understood by the general readership of eLife.

Essential revisions:

1) Reviewers found the analysis that analyzed the phase angle of the resets to be compelling because it has the potential for increasing the impact of the work by linking to theoretical models from rodents. However, several concerns about this analysis were raised:

a) A concern was raised that it is possible that the 180 degree phase difference found between original and updated fixations is caused by the examination of a later relative time offset for fixations to updated positions. If this tricky analysis was understood correctly, it would be important to show that the apparent phase difference they report is truly a result of the resets occurring to different phases rather than just being a result of the analysis measuring phases at different time intervals in each of the two conditions.

b) The plot on the right panel of Figure 2E is very hard to read because the points are small and the blue and green colors are similar. This plot should be made more understandable, perhaps by showing separate histograms of the blue and green distributions as well as by performing a statistical test that is matched to the data (there is a description of some test in the text on this but it is hard to tell if this refers to the exact data in this figure?).

c) Regarding this result, an analysis at the individual electrode level was mentioned. Reviewers had a hard time understanding this analysis and thus it requires clarification. In general, it would be better if they could clarify what this effect means at the individual electrode level, perhaps by showing data from a representative electrode in both conditions. They should also better explain the statistical test in subsection “Theta dependence of memory-guided eye movements” where they mentioned that the effect is significant at 27 of 32 electrodes. It is presumed that they performed some sort of within-electrode, across-trials comparison (which test?). This test was described as "across electrodes" in the preceding sentence (subsection “Theta dependence of memory-guided eye movements”), which may be a mistake.

2) There is a lack of discussion of anatomical specificity of recordings. Specific points related to this are listed below:

a) A concern was raised about the claim that the findings are specific to the human hippocampus. The authors state that they included a mix of depth and grid electrodes and state the number of hippocampal contacts in each subject. These numbers were viewed as unlikely given that there are grid electrodes for one, and in some the intercontact spacing is 5 mm. Having 8 contacts in the hippocampus of a single subject at least with traditional epilepsy electrode implantations is high. Where and how did these 8 contacts come from – depth electrodes, most distal contact? The authors should further detail how these numbers were calculated in the subjects. Were depth electrodes implanted orthogonally? Figure 2A is really not helpful. For bipolar electrodes, how did the authors determine placement; was 1 or both of the contacts in the hippocampus? If indeed the electrode channels are from areas beyond the hippocampus, the authors should revise the claims of their paper accordingly. Furthermore, the number of contacts in Table 3 versus Table 4 does not match.

b) In its current form there is no fair consideration of where in the larger hippocampal formation the observed effects were measured when compared to other studies (in humans, nonhuman primates, and rodents). A discussion of this is necessary to allow interpreting the reported findings with prior signatures.

3) To a reader with knowledge about theta oscillations it is unclear how the reported theta oscillations in the human patients+ electrodes appear (the added waveform shape analysis does not help here). It seems pivotal to show example traces within a e.g. 1-20Hz bandpass signal. Showing these traces is necessary to allow comparison to findings in other studies/species.

4) With only 5 participants, it is possible that some of the results can be driven by a single subject or electrode(s). For example, a peak in theta (4-6 Hz) across the group can be reflected by a small number of electrodes or a few subjects. There seems to be a lot of variability in their group with respect to at least theta from the data that we can see. For specificity of the theta peak in each subject however, there is no way to tell as the paper stands currently. The above point about the larger anatomical concerns could potentially explain some of this variability.

5) What was the 60 second distraction task exactly? Were the subjects given a specific goal or task instruction?

6) A few concerns regarding presentation of statistical analyses were raised:

a) The sample sizes for which statistical analyses are done should be reported, in the text and in each associated figure legend panel. Currently, this is not the case.

b) With regards to Table 2, how was significance here determined?

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Hippocampal theta coordinates memory processing during visual exploration" for further consideration by eLife. Your revised article has been evaluated by Laura Colgin (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

Reviewer #1:The revised manuscript is considerably enhanced and now provides a rich set of important results (including several added control analysis that strengthen the authors main conclusions) and an extended and fair discussion.

The only aspect that I think deserves attention is the description of the PAC results and how they are interpreted:

a) Description of PAC results:

The authors report in subsection “Theta to gamma phase amplitude coupling predicts memory updating” that "[…] 25% of electrodes exhibited significant differences in PAC driven by associative novelty (i.e., differences between fixations to updated or repeated locations)[…]).

The critical question here, however, is not whether there are sign differences in more channels than expected by chance, but how many of them show enhanced PAC for updated locations?

It is confusing (and reflects overstating) to read e.g. subsection “Theta to gamma phase amplitude coupling predicts memory updating” that enhanced PAC when an object-location is forgotten is interpreted in the text as reflecting that memory is updated.

b) Interpretation:

The PAC results are interpreted as indicating not only that enhanced PAC is linked to forgetting (which seems backed up by the results), but that PAC is associated with memory updating. I do not see a result that suggests that a behavioral measure of updated memory is correlated with PAC. Hence this interpretation should not be made or some more explicit and direct result should be added to support that claim.

This affects the Abstract (which states "[…], but predicted memory updating") and it affects the heading of subsection “Theta to gamma phase amplitude coupling predicts memory updating”.

Reviewer #2:

The authors have done a nice job responding to my concern and I think the paper is appropriate for publication. I especially like the new Figure 3, which explains the results rather intuitively.

eLife. 2020 Mar 13;9:e52108. doi: 10.7554/eLife.52108.sa2

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewers all found the topic and experimental design compelling. However, all had significant concerns about the statistics and analyses that raised questions about whether the paper's conclusions were strongly supported by the results. The separate reviews are included below in their entirety, but major concerns that led to the rejection decision include:1) The time-window for analysis: It is unclear that the data only include the fixation of interest (and not any other saccade or fixation). Changing this window would dramatically decrease the available data for analysis to ~250 ms for each fixation.

The results in our manuscript necessarily include multiple behavioral events (eye movements) per time window, given the frequency of eye movements. In our revised manuscript, we provide an additional analysis that shows our findings are consistent when excluding individual fixations with temporally proximal saccades (clean windows ranging from 100 to 400 ms). These results are described in subsection “Theta dependence of memory-guided eye movements” of the text, and displayed in Figure 2—figure supplement 3.

2) Improper statistics: Analyses lacked a direct test of the interaction of interest (and instead concluded differential effects based on the finding that one result was significant and another was not).

We have carefully reviewed the manuscript for circumstances where we implied a statistical difference without explicitly testing for such differences. These circumstances were limited to two occasions, both in the statement regarding subsequent memory effects being specific to viewing of Updated locations on Mismatch trials (subsection “Theta phase consistency during updated-location viewing predicts subsequent memory”) and specificity of our PAC findings to the 4-6 Hz range (subsection “Theta to gamma phase amplitude coupling predicts memory updating”). We now provide statistical tests that directly compare the interaction between theta ITC and subsequent memory during fixations to Updated and Repeated object locations in support of this claim. Figure 3 now depicts subsequent memory effects for both viewing regions of interest, as well as the difference between these two conditions. This comparison is now reported in the main text in subsection “Rapid sequential fixations do not account for memory-specific theta effects”.

In addition, we now use a repeated measures ANOVA to compare PAC effects as a function of frequency band and memory condition (subsection “Theta to gamma phase amplitude coupling predicts memory updating”). In support of this claim, we now report tests showing that interactions between theta phase and gamma amplitude were greatest when using 4 to 6 Hz vs. higher or lower theta frequencies (subsection “Theta to gamma phase amplitude coupling predicts memory updating”).

We would also like to emphasize that the experimental design was not built off a factorial design, where it is straightforward to test for an interaction between conditions. While there are two experimental conditions of interest (Match and Mismatch trials), differences in memory processes only vary within the Mismatch condition. That is, while both novelty and retrieval can guide fixations in the Mismatch condition, comparable viewing states are not present in the Match condition. As a result, an interaction analysis is not practical for the majority of the reported analyses. The comparison to Match trial acts as an important control, demonstrating, for example, that differences in theta phase are not primarily driven by fixating on the presented object.

3) The PAC results were viewed as a weak part of the paper. Concerns were raised that PAC effects may have resulted from changes in power or waveform asymmetry and that analyses were not corrected for multiple comparisons.Differences in the gamma frequency observed (very fast) compared to previous results in rodents (e.g., Colgin et al.) made the discussion of the results as related to slow and fast gamma strange.

We agree with the reviewers that we did not provide strong evidence that our PAC results were not driven by changes in underlying properties of the theta waveform. We now include additional control analyses (subsection “Theta to gamma phase amplitude coupling predicts memory updating”) that demonstrate that while measures of PAC were influenced by theta sharpness and power, they do not account for the observed differences in PAC related to memory. In addition, we have refocused our discussion to similar PAC findings in humans, where appropriate.

4) There was a lack of demonstration of peaks in the spectra for the frequency ranges that were chosen for analysis to justify why these particular frequencies were selected. It is possible that similar results observed at other frequencies and that drawing conclusions about specific oscillatory patterns are unwarranted.

We now demonstrate clear peaks in the power spectra from 4 to 6 Hz across patients and conditions (Figure 2). We note that our focus on theta was driven by previous findings in nonhuman primates (e.g. Jutras et al., 2013) relating hippocampal theta to memory encoding. To provide some evidence regarding the specificity of these results with regard to higher frequency oscillations, we provide an additional exploratory analysis that looks at effects from 1 to 50 Hz, which fails to identify robust effects outside the theta/alpha range (Figure 4—figure supplement 2).

5) Nothing is shown to describe inter-subject variability.

We now characterize the degree to which ITC (and PAC) effects vary across individual subjects. While greater anatomical description of these effects (e.g., comparing across hippocampal subfields or along the anterior-posterior axis) would be ideal, we do not have sufficient coverage across this group of patients to examine variability within different regions of the hippocampus. This information can be found in Table 2, Table 3, Supplementary file 2 and Supplementary file 3.

6) Pre- and post- fixation results were not shown for Match trials (i.e., similar to Figure 2).

We now display the pre- and post-fixation results comparing fixations to individual ROIs on Match and Mismatch trials. Graphical depictions of these comparisons were excluded from the initial manuscript to focus the message to significant findings. The results of this analysis are now graphically depicted in Figure 4—figure supplement 1.

Reviewer #1:[…]- The most substantial concern I see with the paper is that many of the paper's claims hinge on a comparison of two statistical tests, with the authors making much of the fact that one comparison is significant and that another is not. Instead the paper needs to directly test whether one effect is larger than the other, such as by testing the interaction with an ANOVA or some other way. This is a problem in many sections of the paper and substantially undermines the specificity of key claims. The authors should review all their claims and ensure that result is supported by a specific statistical test/interaction, rather than just relying on an effect being significant in one condition and not in another statistically.

We thank the reviewer for bringing this point to our attention. We have carefully reviewed the manuscript and included additional statistical tests, where necessary. See our response to general comment 2 for more details.

- I was disappointed that the paper only measured the magnitude of phase locking, while seemingly ignoring the key issue of what specific phase the locking occurred to. The models that they refer to from Hasselmo and Colgin have strong hypotheses related to specific theta phases and the recorded data would seem to measure this data. But unfortunately, the reported data analyses ignore testing which theta phase the locking occurs to, instead focusing only the magnitude of locking. Without this information, the paper doesn't really provide strong support for those theoretical models.

We agree with the reviewer that strong support for these theoretical models would show phase-specific effects. To test these theories, we identified the preferred phase angles at the onset of each cluster related to novelty and retrieval processes (Figure 2E, left). As shown in the right panel of Figure 2E, we found that phase-locking associated with retrieval began near the peak of theta, whereas novelty-related effects were timed to the trough of theta. A more detailed description of these results can be found in subsection “Theta dependence of memory-guided eye movements”. We thank the reviewer for suggesting this additional analysis, as it directly ties our findings to these theories of hippocampal theta function.

- It is notable that the latencies of the effects are close to 300ms post saccade. Are the effects related to P300s?

We performed additional analysis of ERPs to individual fixations, to examine the possibility that (1) phase locking at different frequencies and (2) ERP components are related to the observed differences in ITC across conditions.

- The PAC analyses are not compelling because they do not rule out the possibility that the effects are driven by changes in power. Based on the data that they show, I think it remains possible that their PAC results are a direct result of changes in theta power or locking without any specific PAC changes. To substantiate their claim that their findings demonstrate a specific role for PAC in memory, the authors need to do much more to specifically show that their effects are caused by interactions between the timing of gamma and theta rather than power or phase changes in just one of the signals. They also should rule out whether the PAC changes could be related to waveform asymmetry.

We agree with the reviewer that it is important to formally test whether interactions between theta and gamma oscillations are responsible for the observed differences in PAC, as opposed to changes in power or underlying theta waveforms. The current version of the manuscript now includes control analyses to rule out the possibility that PAC changes resulted from differences in (1) theta power, (2) theta phase-locking, (3) waveform sharpness, or (4) waveform asymmetry (these results are described in the new subsection “Memory-related changes in PAC are unrelated to theta waveform properties”).

For each significant memory-related PAC effect (i.e., those reported in Figure 5A and 5B), we examined whether underlying properties of the theta waveform predicted concurrent changes in gamma amplitude. While we did find significant (but modest) relationships between multiple theta waveform properties at individual electrodes (and at the group level related to theta power; see Supplementary file 2 and Supplementary file 3), we performed a regression analyses to remove any variance in PAC related to these factors. The results of these analyses, which show consistent memory-related modulates in PAC, are depicted in Figure 6.

In our first analysis, we quantified changes in the LFP evoked by individual fixations to each ROI. Grand average (across electrodes and patients) ERPs are depicted in the following figure. At the group level, we observed modest differences in event related potentials 600 ms after fixations to original vs. updated locations on Mismatch trials (Author response image 1). In general, ERP effects were heterogenous across electrodes and subjects, leading to these results. These results argue against the idea that observed theta effects are driven by evoked responses, such as the hippocampal P300.

Author response image 1. — Shaded regions indicate SEM.

- I had a hard time following the control analyses in the third paragraph of subsection “Theta dependence of memory-guided eye movements”. Can the text here be reworded so the logic is clearer?

We have revised the manuscript in subsection “Theta dependence of memory-guided eye movements” to clarify these control analyses.

- I was surprised to see high-gamma effects at frequencies as high as 130-150 Hz. Were these results corrected for multiple comparisons? In the Discussion section this pattern is compared to the fast and slow gamma oscillations seen in rodents by Colgin et al. However, this signal is so substantially faster in frequency than the signals seen in rodents that this comparison seems far-fetched.

These analyses were corrected for multiple comparisons (accounting for frequencies and angles for phase), but not for the three different task-based comparisons.

We have modified our Discussion section such that we do not directly make inferences regarding the frequency of gamma oscillations. However, we do believe that there is frequency specificity in the observed PAC findings, which supports the general notion that the speed of gamma-band activity may reflect the function of distinct neural circuits/cognitive functions in humans, as they do in other model systems.

- It seems that most of the paper's statistics are performed across subjects (based on the degrees of freedom). This is a useful, conservative approach, which they should better explain this in the results. They used a different approach in subsection “Theta to gamma phase amplitude coupling predicts memory updating”, when they report percentages of hippocampal electrodes that show each effect, which they should explain and justify.

Our electrode-level analyses were meant to convey the degree to which these effects were observable at a given recording site, as opposed to the typical group-level analysis (treating subjects as random effects) used in neuroimaging studies. In response to other reviewers’ concerns, we now uniformly report electrode and subject-level statistics across all analyses (notably, we refer to this statistical decision in subsection “Theta dependence of memory-guided eye movements”, where we refer to new subject and electrode level ITC effects reported in Table 3).

Reviewer #2:[…]1) I have concerns about the time course of the behavior and the analysis windows chosen. Table 1 shows that fixations last ~200-400 ms, which is consistent with a broad literature. However, many of the analyses use a window which surely encompasses more than the fixation of interest. For example, Figure 2B, the significant cluster of ITC occurs begins ~400 ms prior to the fixation. It is likely that this time window includes the fixation 2-back from the fixation of interest. Similarly, the post-fixation effect shown in Figure 2C likely includes the fixation of interest plus the following fixation. It seems that the cleanest approach would be to limit the analysis to the immediately preceding or following fixation. This may have been the approach taken, but it is unclear from the methods. Similarly, it seems likely that multiple fixations within a given ROI occur in succession (as illustrated in Figure 1B). It is not clear how this would have been controlled for in the analysis.

The reviewer is correct that the time-window of interest will necessarily include extra fixation events potentially before and after the fixation indicated at time zero. As such, any analysis of these time periods requires a trade-off between statistical power (including more trials) and potential confounds that can influence theta (additional events occurring in the time window of interest). Given our interest in characterizing theta power and phase, we necessarily require multiple cycles of a theta to obtain reliable estimates, on the order of seconds for lower frequencies. As a result, only examining ‘fixation free’ epochs would drastically reduce the number of observations and statistical power to detect any effects.

To provide evidence that our findings are not driven by sequential sampling behaviors (e.g., a sequence of fixations from the updated to the original object location in the Mismatch condition), we have repeated the main phase-locking analyses with different ‘clean windows’ where no additional fixations occur in either a pre- or post-fixation time period. These analyses replicated the main findings in the manuscript and are reported in subsection “Theta dependence of memory-guided eye movements” and depicted in Figure 2—figure supplement 3. We additionally provide a description of the number of fixations included in each of these analyses in Supplementary file 1.

2) Because the presentation of the data are for the most part, fairly processed, it would be helpful to show an example of the raw LFP and theta phase locking in an individual fixation. It is also very difficult to understand how many fixations were included in each of the analyses.

As mentioned above, we now provide more detailed information regarding the number of fixations that are included in each analysis. The raw hippocampal LFP with individual fixation events for a single Mismatch trial are provided in Figure 1.

3) For the PAC analysis, it is unclear how the example of the representative electrode fits with the population result. The representative electrode showed increased PAC between theta (~5 Hz) and gamma (~140-170 Hz) for fixations to updated vs repeated objects. By contrast, the population effects showed significant PAC at 80-100 Hz. From Figure 5A, it does look like there may be a small increase in PAC in the 140-170 range, which did not reach significance. Is it possible that different PAC effects were observed at different electrode locations?

As noted in subsection “Reset of hippocampal oscillations during memory-guided eye movements is specific to theta”, we observed a great degree of heterogeneity across individual electrodes regarding the frequencies which exhibit PAC. As a result, not every electrode showed a result that typified the group-level analysis. To avoid potential confusion, we have selected a different example electrode for Figure 4, which is more consistent with group-level results.

4) In the Discussion section, the authors report that "inspection of raw traces revealed that the observed differences in PAC resulted from nested oscillations, as opposed to the modulation in the amplitude of sharp waveforms." This is an interesting finding, and it would be helpful to provide examples of these raw traces.

We now include analysis of theta waveform properties (see our response to issue 3 raised by reviewer 1) and have amended our Discussion section to refer to these analyses directly.

5) During the Refresh phase, the subjects performed a recognition memory task, but I couldn't find these results reported. Importantly, it was not clear whether the subjects received feedback on the accuracy of their responses and how their responses on the Refresh phase correlated with accuracy in the final Test phase.

We now include behavioral performance during the Refresh phase (the same/different judgment) and differences in initial and final responses in the behavioral results subsection “Direct brain recordings linked to memory-driven eye movements”. Subjects were not provided feedback regarding the accuracy of their response (Materials and methods section), and performance was highly correlated across phases of the task.

6) In the Materials and methods section, it is not clear whether a distinct background was used for each of the 16 objects or if the same background was used throughout a block. If a different background was used, was each background shown for a 5s familiarization period prior to the Study phase? It was also not clear whether the flashing of the fixation cross was in the location of the object. Was there one study trial per object/scene? Did all subjects perform all 8 blocks?

A unique background was used for each set of 16 objects (each block), and each block used trial-unique objects. All subjects performed all 8 blocks of the task. We now clarify these details in the Materials and methods section and Results section.

Reviewer #3:[…]Beyond these strengths there are several aspects of the analysis and writing that deserve consideration.1) A major concern with this study is the restricted frequency range analyzed which limits interpretation and understanding of memory updating related dynamics. The reader is presented with phase analysis results of the 1-10 Hz range and the 80-200 Hz range. This restriction does not allow to discern how overall (average) and beta and gamma activity relate to memory updating despite prior studies implicating these frequency ranges in related functions. What is the average time-locked average ERP activity around fixations (does that allows to see a reset like behavior)? Are there peri-fixation effects of ITC effects at beta and gamma activities that are related to updating and remembering fixated objects?

2) Were there power spectral peaks discernible at those frequencies at which phases are interpreted? If not, can this be discussed explicitly? What were the shapes of the power spectra? If power peaks at non-existent or at frequencies away from the max-ITC phase effects that would be important to know to constrain interpretation and guide future studies aimed at finding the neural basis of the ITC effects.

We now display power spectra from hippocampal contacts (Figure 2B) which reveal clear spectral peaks in the 4 to 6 Hz range. We also fit power spectra using a combination of periodic and aperiodic components, and assessed statistical significance of spectral peaks (using an F-test) at each electrode in each experimental condition (Table 2). At the group level, we found that more than half of the electrodes exhibited statistically significant oscillations in this frequency range.

3) Figure 2 (legend and main text) insinuate that the analyzed fixations are all "retrieval-dependent". I think this is a misleading statement because it is not made sufficiently clear which fixations were used and only a small fraction may relate to retrieval. Were only Mismatch fixation used? Only from trials which were later 'remembered'? Including the first fixation?If the behavior suggest that there are ~2 sec where refixations to the mismatch locations are predictive to the remembering, then fixations at other time windows are not linked to later remembering. This should be ideally made clear to the reader.

In Figure 2 and in the main text, we refer to fixations to the original object-location on Mismatch trials (when there is no object present at this location) as retrieval-dependent. As shown in Figure 1D, indicated in blue, fixations to the original object location almost exclusively occurred later in the fixation sequence, following viewing of the presented object. We have clarified in the Results section what we mean by retrieval-dependent.

4) Related to the previous point, it would be appreciated to also see and report the pre- and post-fixational ITC for Match trials (e.g. in a figure supplement similar to Figure 2). If you would have a similar 5Hz ITC in those conditions then the interpretation of the pre-fixational mismatch ITC would be different.

We now include these results in Figure 2—figure supplement 3; as reported in the main text, there is no significant difference in ITC between these conditions.

5) It is unclear which "additional" features are possibly explored (given similar fixation durations) that is suggested by this sentence (subsection “Theta dependence of memory-guided eye movements”) : " the number of fixations to the object was reduced on Mismatch relative to Match trials (paired t-test, t4 = -8.8, P = 0.0009, g = -0.6), indicating that subjects explored additional visual features during Mismatch".

We have clarified that we were referring to the background scene.

6) With five subjects it seems necessary to report about the variability and consistency of the observed main findings more explicitly. How many subjects showed a positive memory effect for mismatch fixation trials (at which specific anatomical locations)? In how many subjects was there a predictive theta ITC? In how many subjects was there a post-fixational memory ITC effect?

Our focus on the theta and gamma frequencies was well motivated by prior work in nonhuman primates and humans. Nonetheless, we have performed an additional ERP analysis to examine the extent to which phaselocking may occur at additional frequencies (see Author response image 1. While we did not observe reliable differences in ERPs across patients and electrodes, we examined changes in spectral power of these ERPs at the electrode levels to identify differences in phase alignments across a broader range of frequencies. As shown in Figure 4—figure supplement 2, increases in pre- vs. post-fixation onset ERP power are observed in the high theta/alpha range (8 to 13 Hz) following fixations to objects during both match and mismatch conditions. These results highlight the specificity of our findings to the theta range and provide additional evidence regarding optimal stimulus encoding via resetting of ongoing theta oscillations. These findings are now reported in the subsection “Reset of hippocampal oscillations during memory-guided eye movements is specific to theta”.

Variability across subjects and electrodes is now described in subsection “Theta dependence of memory-guided eye movements” and Table 2.

7) How often were individual objects of the 16 objects shown (eight times?)? Were they shown the same numbers and durations? If no, then please test if this affects the theta ITC effect

All objects used in the study were trial unique. That is to say, each object appeared once (during study, refresh, and test phases) on the background scene that was used for a given block.

8) Were objects shown in the same sequence in each of the eight blocks? If yes, please show that sequence effects are not explaining the memory predictive theta ITC.

Objects were not repeated across blocks. As a result, there is no possibility for sequential learning across blocks.

9) Were the 16 objects shown at 16 random locations or were some shown at the same location? If the latter is the case, please provide test that the number of objects per location is not a confound.

Objects were shown in unique locations.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Essential revisions:1) Reviewers found the analysis that analyzed the phase angle of the resets to be compelling because it has the potential for increasing the impact of the work by linking to theoretical models from rodents. However, several concerns about this analysis were raised:a) A concern was raised that it is possible that the 180 degree phase difference found between original and updated fixations is caused by the examination of a later relative time offset for fixations to updated positions. If this tricky analysis was understood correctly, it would be important to show that the apparent phase difference they report is truly a result of the resets occurring to different phases rather than just being a result of the analysis measuring phases at different time intervals in each of the two conditions.

We agree with the reviewers that it is important to demonstrate that differences in theta phase related to retrieval and associative novelty do not directly result from differences in the timing of these two effects, relative to fixation onsets. We now present a comparison of phase distributions considering both the viewing region of interest (i.e., updated or original object-location) and time interval (i.e., pre or post fixation onset) to rule out this possibility (see Figure 3). We find two pieces of evidence that support our conclusions. First, in the interval preceding fixations to updated object-locations, there is no evidence that the phase distribution is different from a uniform distribution (Z = 1.04, P = 0.36, Rayleigh test). That is, theta phase is not consistently aligned to a specific angle in this condition, prior to execution of the saccade. Second, following fixations to the original object location, we found theta phase distributions concentrated around the trough of the oscillation (mean = 2.9 rad [2.5-3.4 95% CI]). These phase angles were not distinguishable from those following fixations to the updated object-location (F(62) = 0.77, P = 0.38). Thus, theta phase following fixations on Mismatch trials are generally aligned to the trough; however, alignment to the peak of theta only occurs when we believe retrieval processes are necessary (preceding fixations to the original object-location). We have added a statement clarifying this point in the Results section.

b) The plot on the right panel of Figure 2E is very hard to read because the points are small and the blue and green colors are similar. This plot should be made more understandable, perhaps by showing separate histograms of the blue and green distributions as well as by performing a statistical test that is matched to the data (there is a description of some test in the text on this but it is hard to tell if this refers to the exact data in this figure?).

We agree that the results presented in Figure 2E were difficult to read. To more prominently display these findings, we have added an additional figure (Figure 3) that presents these results both in the form of polar histograms and in addition to the previous format (which we believe clearly communicates the peak/trough relationship to reader unfamiliar with polar plots). We have also rewritten the text in the Results section to precisely follow what is depicted in the figure.

c) Regarding this result, an analysis at the individual electrode level was mentioned. Reviewers had a hard time understanding this analysis and thus it requires clarification. In general, it would be better if they could clarify what this effect means at the individual electrode level, perhaps by showing data from a representative electrode in both conditions. They should also better explain the statistical test in subsection “Theta dependence of memory-guided eye movements” where they mentioned that the effect is significant at 27 of 32 electrodes. It is presumed that they performed some sort of within-electrode, across-trials comparison (which test?). This test was described as "across electrodes" in the preceding sentence (subsection “Theta dependence of memory-guided eye movements”), which may be a mistake.

In order to make this analysis more intuitive to readers, we now describe the procedure in a descriptive manner from measuring theta phase relative to specific fixations (Figure 3B), to comparing differences at the level of individual electrodes (Figure 3C), and finally across all electrodes (Figure 3D). We also thank the reviewers for catching an error introduced during editing, which has been corrected.

2) There is a lack of discussion of anatomical specificity of recordings. Specific points related to this are listed below:a) A concern was raised about the claim that the findings are specific to the human hippocampus. The authors state that they included a mix of depth and grid electrodes and state the number of hippocampal contacts in each subject. These numbers were viewed as unlikely given that there are grid electrodes for one, and in some the intercontact spacing is 5 mm. Having 8 contacts in the hippocampus of a single subject at least with traditional epilepsy electrode implantations is high. Where and how did these 8 contacts come from – depth electrodes, most distal contact? The authors should further detail how these numbers were calculated in the subjects. Were depth electrodes implanted orthogonally? Figure 2A is really not helpful. For bipolar electrodes, how did the authors determine placement; was 1 or both of the contacts in the hippocampus? If indeed the electrode channels are from areas beyond the hippocampus, the authors should revise the claims of their paper accordingly. Furthermore, the number of contacts in table 3 versus 4 does not match.

We agree with the reviewers that the number of contacts is high for typical electrode implantations with lateral trajectories targeting the hippocampus. We now clarify two points in the methods: (1) while patients were implanted with subdural grids, all analyses focused on depths targeting hippocampus, and (2) following bipolar referencing, all pairs with at least one contact within grey matter or proximal tissue were analyzed. This led to our inclusion of 31 contacts (yielding 32 bipolar pairs, accounting for the discrepancy between Table 3 and Table 4). In addition, we have updated Figure 2A to clarify the position of contact locations within each patient within the hippocampus. Even though recordings were lateralized, increased hippocampal coverage resulted from multiple hippocampal depth electrodes per patient.

b) In its current form there is no fair consideration of where in the larger hippocampal formation the observed effects were measured when compared to other studies (in humans, nonhuman primates, and rodents). A discussion of this is necessary to allow interpreting the reported findings with prior signatures.

In our previous iteration of the manuscript, we omitted discussion of anatomical localization within the hippocampus as our coverage was limited to the head and body of the hippocampus. We have expanded our discussion to include comparisons of our results to other studies, with specific focus on understanding functional organization across the long axis of the hippocampus. We also emphasize limitations of our dataset with regards to the anatomical specificity of our findings, as we did not obtain recordings in posterior hippocampus or resolve difference within hippocampal subfields.

3) To a reader with knowledge about theta oscillations it is unclear how the reported theta oscillations in the human patients+ electrodes appear (the added waveform shape analysis does not help here). It seems pivotal to show example traces within a e.g. 1-20Hz bandpass signal. Showing these traces is necessary to allow comparison to findings in other studies/species.

We now provide additional examples of raw traces, both filtered below 20 Hz and within the predominant theta frequency (4-6 Hz) in Figure 3 of the main text. To provide additional examples that may highlight anatomical variability, additional traces are shown in Figure 2—figure supplement 1.

4) With only 5 participants, it is possible that some of the results can be driven by a single subject or electrode(s). For example, a peak in theta (4-6 Hz) across the group can be reflected by a small number of electrodes or a few subjects. There seems to be a lot of variability in their group with respect to at least theta from the data that we can see. For specificity of the theta peak in each subject however, there is no way to tell as the paper stands currently. The above point about the larger anatomical concerns could potentially explain some of this variability.

The fact that electrode and subject effects can drive grand averages is precisely the reason we evaluated the presence of 4-6 Hz oscillations at the individual electrode level. As reported in Table 2, nearly 80% of electrodes exhibited statistically significant oscillations at this frequency irrespective of condition, except for after fixations to the original object-location (which did not involve viewing of a presented stimulus) in which only 50% of bipolar pairs exhibited significant oscillatory activity. We now present power spectra for individual electrodes in Figure 2—figure supplement 1, which demonstrates consistency of this effect.

5) What was the 60 second distraction task exactly? Were the subjects given a specific goal or task instruction?

During the distraction task, subjects performed free-viewing of unrelated visual scenes (cats). The stimulus pool and task are now described in the Materials and methods section.

6) A few concerns regarding presentation of statistical analyses were raised:a) The sample sizes for which statistical analyses are done should be reported, in the text and in each associated figure legend panel. Currently, this is not the case.

We now specify sample sizes for all tests and summary measures.

b) With regards to Table 2, how was significance here determined?

To determine whether statistically significant oscillations were present at each electrode, we modeled the power spectra for each period of interest (i.e., each epoch around a given fixation) as a mixture of aperiodic (modeled as 1/f) and periodic (modeled as gaussians) components. After model fitting, we performed a test of equal variance (across fixations) between the raw power spectra and the power spectra after removing modeled oscillations, yielding an F-statistic. If oscillations were present in the data, one would expect unequal variances across these two distributions. Statistical significance for each electrode was assessed using a permutation procedure (n = 10,000), randomizing assignment of psd-type (i.e., raw or aperiodic) across fixations. Statistical tests at the subject and group level were performed using binomial tests, comparing the proportion of significant electrodes to chance (established as 0.05 in the permutation procedure), with an α of 0.05, Bonferroni corrected for multiple comparisons across conditions and time periods (i.e., pre- post-fixation).

This description of model fitting is explained in the subsection “Spectral decomposition”. We have added description of the statistical procedure to the subsection “Statistical analyses”.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #1:The revised manuscript is considerably enhanced and now provides a rich set of important results (including several added control analysis that strengthen the authors main conclusions) and an extended and fair discussion.The only aspect that I think deserves attention is the description of the PAC results and how they are interpreted:a) Description of PAC results:The authors report in subsection “Theta to gamma phase amplitude coupling predicts memory updating” that "[…] 25% of electrodes exhibited significant differences in PAC driven by associative novelty (i.e., differences between fixations to updated or repeated locations)[…]).The critical question here, however, is not whether there are sign differences in more channels than expected by chance, but how many of them show enhanced PAC for updated locations?It is confusing (and reflects overstating) to read e.g. subsection “Theta to gamma phase amplitude coupling predicts memory updating” that enhanced PAC when an object-location is forgotten is interpreted in the text as reflecting that memory is updated.

In this analysis, we were primarily concerned with whether PAC was modulated by associative novelty, rather than focusing on absolute levels of PAC within a given condition (e.g., comparing PAC during fixations to updated locations against a reasonable null, such as surrogates constructed by permuting phase information across fixations). We agree that testing for differences in PAC during the viewing of updated locations would provide additional evidence that PAC is related to memory updating. We now report the suggested analysis (subsection “Theta to gamma phase amplitude coupling predicts memory updating”), comparing PAC following fixations to updated object-locations to phase-permuted null distributions. We found that 16% of hippocampal electrodes (significantly more than expected by chance, P = 0.02, binomial test) exhibited increased theta to gamma PAC. Based on these results and our analysis of forgetting-related changes in PAC, we do not believe we are overstating our findings.

b) Interpretation:The PAC results are interpreted as indicating not only that enhanced PAC is linked to forgetting (which seems backed up by the results), but that PAC is associated with memory updating. I do not see a result that suggests that a behavioral measure of updated memory is correlated with PAC. Hence this interpretation should not be made or some more explicit and direct result should be added to support that claim.This affects the Abstract (which states "[…], but predicted memory updating") and it affects the heading of subsection “Theta to gamma phase amplitude coupling predicts memory updating”.

Our claim that increased PAC is associated with memory updating is based on both (1) greater PAC during fixations to updated vs. repeated object-locations and (2) greater PAC during fixations to updated object-locations when the original location is forgotten (vs. remembered). This inference does rely on behavior on the task. When subjects forgot the original object-location on mismatch trials, they reported the updated object-location on 73% of trials. That is to say, interference from the refresh phase disrupted memory for the original location. We now clarify this point when presenting the results (subsection “Theta to gamma phase amplitude coupling predicts memory updating”). In our discussion of these results (Discussion section), we now emphasize that forgetting on the task was primarily due to updating of the object location. This subtlety is important when considering the role of hippocampal PAC in memory formation, which is commonly evaluated during initial encoding.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

Kragel JE, VanHaerents S, Templer JW, Schuele S, Rosenow JM, Nilakantan AS, Bridge DJ. 2019. Simultaneous eye tracking and hippocampal iEEG to identify oscillatory signals of memory retrieval and novelty detection in humans. NIMH Data Archive. 2890

Supplementary Materials

Figure 1—source data 1. MATLAB code and source files to reproduce data in Figure 1.

elife-52108-fig1-data1.zip^{(10.1KB, zip)}

Figure 2—source data 1. MATLAB code and source files to reproduce data in Figure 2.

elife-52108-fig2-data1.zip^{(1.2MB, zip)}

Figure 2—figure supplement 1—source data 1. MATLAB code and source files to reproduce data in Figure 2—figure supplement 1.

elife-52108-fig2-figsupp1-data1.zip^{(1.2MB, zip)}

Figure 2—figure supplement 2—source data 1. MATLAB code and source files to reproduce data in Figure 2—figure supplement 2.

elife-52108-fig2-figsupp2-data1.zip^{(271.1KB, zip)}

Figure 2—figure supplement 3—source data 1. MATLAB code and source files to reproduce data in Figure 2—figure supplement 3.

elife-52108-fig2-figsupp3-data1.zip^{(809.1KB, zip)}

Figure 3—source data 1. MATLAB code and source files to reproduce data in Figure 3.

elife-52108-fig3-data1.zip^{(11.5MB, zip)}

Figure 4—source data 1. MATLAB code and source files to reproduce data in Figure 4.

elife-52108-fig4-data1.zip^{(1.4MB, zip)}

Figure 4—figure supplement 1—source data 1. MATLAB code and source files to reproduce data in Figure 4—figure supplement 1.

elife-52108-fig4-figsupp1-data1.zip^{(2.4MB, zip)}

Figure 4—figure supplement 2—source data 1. MATLAB code and source files to reproduce data in Figure 4—figure supplement 2.

elife-52108-fig4-figsupp2-data1.zip^{(2.8MB, zip)}

Figure 5—source data 1. MATLAB code and source files to reproduce data in Figure 5.

elife-52108-fig5-data1.zip^{(75.9MB, zip)}

Figure 6—source data 1. MATLAB code and source files to reproduce data in Figure 6.

elife-52108-fig6-data1.zip^{(17.8MB, zip)}

Supplementary file 1. Number of fixations for each comparison of interest.

elife-52108-supp1.docx^{(16.9KB, docx)}

Transparent reporting form

elife-52108-transrepform.docx^{(245.9KB, docx)}

Data Availability Statement

Behavioral data, eye movement data, continuous EEG recordings, and electrode locations in MNI space have been uploaded to NDAR.

The following dataset was generated:

[bib1] Aru J, Aru J, Priesemann V, Wibral M, Lana L, Pipa G, Singer W, Vicente R. Untangling cross-frequency coupling in neuroscience. Current Opinion in Neurobiology. 2015;31:51–61. doi: 10.1016/j.conb.2014.08.002. [DOI] [PubMed] [Google Scholar]

[bib2] Axmacher N, Elger CE, Fell J. Ripples in the medial temporal lobe are relevant for human memory consolidation. Brain. 2008;131:1806–1817. doi: 10.1093/brain/awn103. [DOI] [PubMed] [Google Scholar]

[bib3] Axmacher N, Henseler MM, Jensen O, Weinreich I, Elger CE, Fell J. Cross-frequency coupling supports multi-item working memory in the human hippocampus. PNAS. 2010;107:3228–3233. doi: 10.1073/pnas.0911531107. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Bellmund JLS, Gärdenfors P, Moser EI, Doeller CF. Navigating cognition: spatial codes for human thinking. Science. 2018;362:eaat6766. doi: 10.1126/science.aat6766. [DOI] [PubMed] [Google Scholar]

[bib5] Berens P. CircStat: a MATLAB toolbox for circular statistics. Journal of Statistical Software. 2009;31:1–21. doi: 10.18637/jss.v031.i10. [DOI] [Google Scholar]

[bib6] Bicanski A, Burgess N. A computational model of visual recognition memory via grid cells. Current Biology. 2019;29:979–990. doi: 10.1016/j.cub.2019.01.077. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib7] Bragin A, Jandó G, Nádasdy Z, Hetke J, Wise K, Buzsáki G. Gamma (40-100 hz) oscillation in the hippocampus of the behaving rat. The Journal of Neuroscience. 1995;15:47–60. doi: 10.1523/JNEUROSCI.15-01-00047.1995. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib8] Bridge DJ, Cohen NJ, Voss JL. Distinct hippocampal versus frontoparietal network contributions to retrieval and Memory-guided Exploration. Journal of Cognitive Neuroscience. 2017;29:1324–1338. doi: 10.1162/jocn_a_01143. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] Bridge DJ, Paller KA. Neural correlates of reactivation and retrieval-induced distortion. Journal of Neuroscience. 2012;32:12144–12151. doi: 10.1523/JNEUROSCI.1378-12.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib10] Bridge DJ, Voss JL. Active retrieval facilitates across-episode binding by modulating the content of memory. Neuropsychologia. 2014a;63:154–164. doi: 10.1016/j.neuropsychologia.2014.08.024. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] Bridge DJ, Voss JL. Hippocampal binding of novel information with dominant memory traces can support both memory stability and change. Journal of Neuroscience. 2014b;34:2203–2213. doi: 10.1523/JNEUROSCI.3819-13.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib12] Bridge DJ, Voss JL. Binding among select episodic elements is altered via active short-term retrieval. Learning & Memory. 2015;22:360–363. doi: 10.1101/lm.038703.115. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib13] Brodeur MB, Dionne-Dostie E, Montreuil T, Lepage M. The bank of standardized stimuli (BOSS), a new set of 480 normative photos of objects to be used as visual stimuli in cognitive research. PLOS ONE. 2010;5:e10773. doi: 10.1371/journal.pone.0010773. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] Burke JF, Zaghloul KA, Jacobs J, Williams RB, Sperling MR, Sharan AD, Kahana MJ. Synchronous and asynchronous theta and gamma activity during episodic memory formation. Journal of Neuroscience. 2013;33:292–304. doi: 10.1523/JNEUROSCI.2057-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] Buzsáki G, Moser EI. Memory, navigation and theta rhythm in the hippocampal-entorhinal system. Nature Neuroscience. 2013;16:130–138. doi: 10.1038/nn.3304. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] Chen J, Dastjerdi M, Foster BL, LaRocque KF, Rauschecker AM, Parvizi J, Wagner AD. Human hippocampal increases in low-frequency power during associative prediction violations. Neuropsychologia. 2013;51:2344–2351. doi: 10.1016/j.neuropsychologia.2013.03.019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib17] Cole SR, van der Meij R, Peterson EJ, de Hemptinne C, Starr PA, Voytek B. Nonsinusoidal beta oscillations reflect cortical pathophysiology in Parkinson's Disease. The Journal of Neuroscience. 2017;37:4830–4840. doi: 10.1523/JNEUROSCI.2208-16.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib18] Cole SR, Voytek B. Brain oscillations and the importance of waveform shape. Trends in Cognitive Sciences. 2017;21:137–149. doi: 10.1016/j.tics.2016.12.008. [DOI] [PubMed] [Google Scholar]

[bib19] Colgin LL. Do slow and fast gamma rhythms correspond to distinct functional states in the hippocampal network? Brain Research. 2015;1621:309–315. doi: 10.1016/j.brainres.2015.01.005. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] Colgin LL. Rhythms of the hippocampal network. Nature Reviews Neuroscience. 2016;17:239–249. doi: 10.1038/nrn.2016.21. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib21] Dahnke R, Yotter RA, Gaser C. Cortical thickness and central surface estimation. NeuroImage. 2013;65:336–348. doi: 10.1016/j.neuroimage.2012.09.050. [DOI] [PubMed] [Google Scholar]

[bib22] Douchamps V, Jeewajee A, Blundell P, Burgess N, Lever C. Evidence for encoding versus retrieval scheduling in the hippocampus by theta phase and acetylcholine. Journal of Neuroscience. 2013;33:8689–8704. doi: 10.1523/JNEUROSCI.4483-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Duncan K, Curtis C, Davachi L. Distinct memory signatures in the hippocampus: intentional states distinguish match and mismatch enhancement signals. Journal of Neuroscience. 2009;29:131–139. doi: 10.1523/JNEUROSCI.2998-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib24] Duncan K, Ketz N, Inati SJ, Davachi L. Evidence for area CA1 as a match/mismatch detector: a high-resolution fMRI study of the human hippocampus. Hippocampus. 2012;22:389–398. doi: 10.1002/hipo.20933. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib25] Dykstra AR, Chan AM, Quinn BT, Zepeda R, Keller CJ, Cormier J, Madsen JR, Eskandar EN, Cash SS. Individualized localization and cortical surface-based registration of intracranial electrodes. NeuroImage. 2012;59:3563–3570. doi: 10.1016/j.neuroimage.2011.11.046. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib26] Fanselow MS, Dong HW. Are the dorsal and ventral hippocampus functionally distinct structures? Neuron. 2010;65:7–19. doi: 10.1016/j.neuron.2009.11.031. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] Haller M, Donoghue T, Peterson E, Varma P, Sebastian P, Gao R, Noto T, Knight RT, Shestyuk A, Voytek B. Parameterizing neural power spectra. bioRxiv. 2018 doi: 10.1101/299859. [DOI] [PMC free article] [PubMed]

[bib28] Hannula DE, Ryan JD, Tranel D, Cohen NJ. Rapid onset relational memory effects are evident in eye movement behavior, but not in hippocampal amnesia. Journal of Cognitive Neuroscience. 2007;19:1690–1705. doi: 10.1162/jocn.2007.19.10.1690. [DOI] [PubMed] [Google Scholar]

[bib29] Hannula DE, Ranganath C. The eyes have it: hippocampal activity predicts expression of memory in eye movements. Neuron. 2009;63:592–599. doi: 10.1016/j.neuron.2009.08.025. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib30] Hasselmo ME, Bodelón C, Wyble BP. A proposed function for hippocampal theta rhythm: separate phases of encoding and retrieval enhance reversal of prior learning. Neural Computation. 2002;14:793–817. doi: 10.1162/089976602317318965. [DOI] [PubMed] [Google Scholar]

[bib31] Hasselmo ME, Eichenbaum H. Hippocampal mechanisms for the context-dependent retrieval of episodes. Neural Networks. 2005;18:1172–1190. doi: 10.1016/j.neunet.2005.08.007. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib32] Hentschke H, Stüttgen MC. Computation of measures of effect size for neuroscience data sets. European Journal of Neuroscience. 2011;34:1887–1894. doi: 10.1111/j.1460-9568.2011.07902.x. [DOI] [PubMed] [Google Scholar]

[bib33] Heusser AC, Poeppel D, Ezzyat Y, Davachi L. Episodic sequence memory is supported by a theta-gamma phase code. Nature Neuroscience. 2016;19:1374–1380. doi: 10.1038/nn.4374. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib34] Hoffman KL, Dragan MC, Leonard TK, Micheli C, Montefusco-Siegmund R, Valiante TA. Saccades during visual exploration align hippocampal 3-8 hz rhythms in human and non-human primates. Frontiers in Systems Neuroscience. 2013;7:43. doi: 10.3389/fnsys.2013.00043. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib35] Honey RC, Watt A, Good M. Hippocampal lesions disrupt an associative mismatch process. The Journal of Neuroscience. 1998;18:2226–2230. doi: 10.1523/JNEUROSCI.18-06-02226.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] Howard LR, Kumaran D, Olafsdottir HF, Spiers HJ. Double dissociation between hippocampal and parahippocampal responses to object-background context and scene novelty. Journal of Neuroscience. 2011;31:5253–5261. doi: 10.1523/JNEUROSCI.6055-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib37] Jacobs J. Hippocampal theta oscillations are slower in humans than in rodents: implications for models of spatial navigation and memory. Philosophical Transactions of the Royal Society B: Biological Sciences. 2014;369:20130304. doi: 10.1098/rstb.2013.0304. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib38] Julian JB, Keinath AT, Frazzetta G, Epstein RA. Human entorhinal cortex represents visual space using a boundary-anchored grid. Nature Neuroscience. 2018;21:191–194. doi: 10.1038/s41593-017-0049-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib39] Jutras MJ, Fries P, Buffalo EA. Oscillatory activity in the monkey hippocampus during visual exploration and memory formation. PNAS. 2013;110:13144–13149. doi: 10.1073/pnas.1302351110. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib40] Kahana MJ, Seelig D, Madsen JR. Theta returns. Current Opinion in Neurobiology. 2001;11:739–744. doi: 10.1016/S0959-4388(01)00278-1. [DOI] [PubMed] [Google Scholar]

[bib41] Killian NJ, Jutras MJ, Buffalo EA. A map of visual space in the primate entorhinal cortex. Nature. 2012;491:761–764. doi: 10.1038/nature11587. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] Kuhl BA, Bainbridge WA, Chun MM. Neural reactivation reveals mechanisms for updating memory. Journal of Neuroscience. 2012;32:3453–3461. doi: 10.1523/JNEUROSCI.5846-11.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib43] Kullback S, Leibler RA. On information and sufficiency. The Annals of Mathematical Statistics. 1951;22:79–86. doi: 10.1214/aoms/1177729694. [DOI] [Google Scholar]

[bib44] Kumaran D, Maguire EA. An unexpected sequence of events: mismatch detection in the human Hippocampus. PloS Biology. 2006;4:e424. doi: 10.1371/journal.pbio.0040424. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib45] Kumaran D, Maguire EA. Match mismatch processes underlie human hippocampal responses to associative novelty. Journal of Neuroscience. 2007a;27:8517–8524. doi: 10.1523/JNEUROSCI.1677-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib46] Kumaran D, Maguire EA. Which computational mechanisms operate in the Hippocampus during novelty detection? Hippocampus. 2007b;17:735–748. doi: 10.1002/hipo.20326. [DOI] [PubMed] [Google Scholar]

[bib47] Kumaran D, Maguire EA. Novelty signals: a window into hippocampal information processing. Trends in Cognitive Sciences. 2009;13:47–54. doi: 10.1016/j.tics.2008.11.004. [DOI] [PubMed] [Google Scholar]

[bib48] Landau AN, Schreyer HM, van Pelt S, Fries P. Distributed attention is implemented through theta-rhythmic gamma modulation. Current Biology. 2015;25:2332–2337. doi: 10.1016/j.cub.2015.07.048. [DOI] [PubMed] [Google Scholar]

[bib49] Lega BC, Jacobs J, Kahana M. Human hippocampal theta oscillations and the formation of episodic memories. Hippocampus. 2012;22:748–761. doi: 10.1002/hipo.20937. [DOI] [PubMed] [Google Scholar]

[bib50] Lega B, Burke J, Jacobs J, Kahana MJ. Slow-theta-to-gamma phase-amplitude coupling in human hippocampus supports the formation of new episodic memories. Cerebral Cortex. 2016;26:268–278. doi: 10.1093/cercor/bhu232. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib51] Leonard TK, Mikkila JM, Eskandar EN, Gerrard JL, Kaping D, Patel SR, Womelsdorf T, Hoffman KL. Sharp wave ripples during visual exploration in the primate hippocampus. Journal of Neuroscience. 2015;35:14771–14782. doi: 10.1523/JNEUROSCI.0864-15.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib52] Leonard TK, Hoffman KL. Sharp-wave ripples in primates are enhanced near remembered visual objects. Current Biology. 2017;27:257–262. doi: 10.1016/j.cub.2016.11.027. [DOI] [PubMed] [Google Scholar]

[bib53] Lever C, Burton S, Jeewajee A, Wills TJ, Cacucci F, Burgess N, O'Keefe J. Environmental novelty elicits a later theta phase of firing in CA1 but not subiculum. Hippocampus. 2010;20:229–234. doi: 10.1002/hipo.20671. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib54] Lisman JE. Relating hippocampal circuitry to function: recall of memory sequences by reciprocal dentate-CA3 interactions. Neuron. 1999;22:233–242. doi: 10.1016/s0896-6273(00)81085-5. [DOI] [PubMed] [Google Scholar]

[bib55] Lisman JE, Grace AA. The hippocampal-VTA loop: controlling the entry of information into long-term memory. Neuron. 2005;46:703–713. doi: 10.1016/j.neuron.2005.05.002. [DOI] [PubMed] [Google Scholar]

[bib56] Lisman JE, Jensen O. The theta-gamma neural code. Neuron. 2013;77:1002–1016. doi: 10.1016/j.neuron.2013.03.007. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib57] Liu ZX, Shen K, Olsen RK, Ryan JD. Visual sampling predicts hippocampal activity. The Journal of Neuroscience. 2017;37:599–609. doi: 10.1523/JNEUROSCI.2610-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib58] Long NM, Lee H, Kuhl BA. Hippocampal mismatch signals are modulated by the strength of neural predictions and their similarity to outcomes. The Journal of Neuroscience. 2016;36:12677–12687. doi: 10.1523/JNEUROSCI.1850-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib59] Lopes-Dos-Santos V, van de Ven GM, Morley A, Trouche S, Campo-Urriza N, Dupret D. Parsing hippocampal theta oscillations by nested spectral components during spatial exploration and memory-guided behavior. Neuron. 2018;100:940–952. doi: 10.1016/j.neuron.2018.09.031. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib60] Lucas HD, Duff MC, Cohen NJ. The hippocampus promotes effective saccadic information gathering in humans. Journal of Cognitive Neuroscience. 2019;31:186–201. doi: 10.1162/jocn_a_01336. [DOI] [PubMed] [Google Scholar]

[bib61] Manns JR, Zilli EA, Ong KC, Hasselmo ME, Eichenbaum H. Hippocampal CA1 spiking during encoding and retrieval: relation to theta phase. Neurobiology of Learning and Memory. 2007;87:9–20. doi: 10.1016/j.nlm.2006.05.007. [DOI] [PubMed] [Google Scholar]

[bib62] Maris E, Oostenveld R. Nonparametric statistical testing of EEG- and MEG-data. Journal of Neuroscience Methods. 2007;164:177–190. doi: 10.1016/j.jneumeth.2007.03.024. [DOI] [PubMed] [Google Scholar]

[bib63] Meister MLR, Buffalo EA. Getting directions from the hippocampus: the neural connection between looking and memory. Neurobiology of Learning and Memory. 2016;134:135–144. doi: 10.1016/j.nlm.2015.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib64] Miller J, Watrous AJ, Tsitsiklis M, Lee SA, Sheth SA, Schevon CA, Smith EH, Sperling MR, Sharan A, Asadi-Pooya AA, Worrell GA, Meisenhelter S, Inman CS, Davis KA, Lega B, Wanda PA, Das SR, Stein JM, Gorniak R, Jacobs J. Lateralized hippocampal oscillations underlie distinct aspects of human spatial memory and navigation. Nature Communications. 2018;9:2423. doi: 10.1038/s41467-018-04847-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib65] Mormann F, Fell J, Axmacher N, Weber B, Lehnertz K, Elger CE, Fernández G. Phase/amplitude reset and theta-gamma interaction in the human medial temporal lobe during a continuous word recognition memory task. Hippocampus. 2005;15:890–900. doi: 10.1002/hipo.20117. [DOI] [PubMed] [Google Scholar]

[bib66] Mormann F, Osterhage H, Andrzejak RG, Weber B, Fernández G, Fell J, Elger CE, Lehnertz K. Independent Delta/theta rhythms in the human hippocampus and entorhinal cortex. Frontiers in Human Neuroscience. 2008;2:3. doi: 10.3389/neuro.09.003.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib67] Morris RG, Inglis J, Ainge JA, Olverman HJ, Tulloch J, Dudai Y, Kelly PA. Memory reconsolidation: sensitivity of spatial memory to inhibition of protein synthesis in dorsal hippocampus during encoding and retrieval. Neuron. 2006;50:479–489. doi: 10.1016/j.neuron.2006.04.012. [DOI] [PubMed] [Google Scholar]

[bib68] Nau M, Navarro Schröder T, Bellmund JLS, Doeller CF. Hexadirectional coding of visual space in human entorhinal cortex. Nature Neuroscience. 2018;21:188–190. doi: 10.1038/s41593-017-0050-8. [DOI] [PubMed] [Google Scholar]

[bib69] Newman EL, Gillet SN, Climer JR, Hasselmo ME. Cholinergic blockade reduces theta-gamma phase amplitude coupling and speed modulation of theta frequency consistent with behavioral effects on encoding. Journal of Neuroscience. 2013;33:19635–19646. doi: 10.1523/JNEUROSCI.2586-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib70] Olsen RK, Sebanayagam V, Lee Y, Moscovitch M, Grady CL, Rosenbaum RS, Ryan JD. The relationship between eye movements and subsequent recognition: evidence from individual differences and amnesia. Cortex. 2016;85:182–193. doi: 10.1016/j.cortex.2016.10.007. [DOI] [PubMed] [Google Scholar]

[bib71] Poppenk J, Evensmoen HR, Moscovitch M, Nadel L. Long-axis specialization of the human hippocampus. Trends in Cognitive Sciences. 2013;17:230–240. doi: 10.1016/j.tics.2013.03.005. [DOI] [PubMed] [Google Scholar]

[bib72] Re D, Inbar M, Richter CG, Landau AN. Feature-based attention samples stimuli rhythmically. Current Biology. 2019;29:693–699. doi: 10.1016/j.cub.2019.01.010. [DOI] [PubMed] [Google Scholar]

[bib73] Rennó-Costa C, Tort ABL. Place and grid cells in a loop: implications for memory function and spatial coding. The Journal of Neuroscience. 2017;37:8062–8076. doi: 10.1523/JNEUROSCI.3490-16.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib74] Ryan JD, Althoff RR, Whitlow S, Cohen NJ. Amnesia is a deficit in relational memory. Psychological Science. 2000;11:454–461. doi: 10.1111/1467-9280.00288. [DOI] [PubMed] [Google Scholar]

[bib75] Scheffer-Teixeira R, Tort AB. On cross-frequency phase-phase coupling between theta and gamma oscillations in the hippocampus. eLife. 2016;5:e20515. doi: 10.7554/eLife.20515. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib76] Sederberg PB, Kahana MJ, Howard MW, Donner EJ, Madsen JR. Theta and gamma oscillations during encoding predict subsequent recall. The Journal of Neuroscience. 2003;23:10809–10814. doi: 10.1523/JNEUROSCI.23-34-10809.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib77] Shirhatti V, Borthakur A, Ray S. Effect of reference scheme on power and phase of the local field potential. Neural Computation. 2016;28:882–913. doi: 10.1162/NECO_a_00827. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib78] Siegle JH, Wilson MA. Enhancement of encoding and retrieval functions through theta phase-specific manipulation of hippocampus. eLife. 2014;3:e03061. doi: 10.7554/eLife.03061. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib79] Smith CN, Hopkins RO, Squire LR. Experience-dependent eye movements, awareness, and hippocampus-dependent memory. Journal of Neuroscience. 2006;26:11304–11312. doi: 10.1523/JNEUROSCI.3071-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib80] Staresina BP, Bergmann TO, Bonnefond M, van der Meij R, Jensen O, Deuker L, Elger CE, Axmacher N, Fell J. Hierarchical nesting of slow oscillations, spindles and ripples in the human hippocampus during sleep. Nature Neuroscience. 2015;18:1679–1686. doi: 10.1038/nn.4119. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib81] Staudigl T, Hartl E, Noachtar S, Doeller CF, Jensen O. Saccades are phase-locked to alpha oscillations in the occipital and medial temporal lobe during successful memory encoding. PloS Biology. 2017;15:e2003404. doi: 10.1371/journal.pbio.2003404. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib82] Strange BA, Witter MP, Lein ES, Moser EI. Functional organization of the hippocampal longitudinal axis. Nature Reviews Neuroscience. 2014;15:655–669. doi: 10.1038/nrn3785. [DOI] [PubMed] [Google Scholar]

[bib83] Thakral PP, Yu SS, Rugg MD. The hippocampus is sensitive to the mismatch in novelty between items and their contexts. Brain Research. 2015;1602:144–152. doi: 10.1016/j.brainres.2015.01.033. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib84] Tort AB, Komorowski RW, Manns JR, Kopell NJ, Eichenbaum H. Theta-gamma coupling increases during the learning of item-context associations. PNAS. 2009;106:20942–20947. doi: 10.1073/pnas.0911331106. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib85] Tort AB, Komorowski R, Eichenbaum H, Kopell N. Measuring phase-amplitude coupling between neuronal oscillations of different frequencies. Journal of Neurophysiology. 2010;104:1195–1210. doi: 10.1152/jn.00106.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib86] Vaz AP, Yaffe RB, Wittig JH, Inati SK, Zaghloul KA. Dual origins of measured phase-amplitude coupling reveal distinct neural mechanisms underlying episodic memory in the human cortex. NeuroImage. 2017;148:148–159. doi: 10.1016/j.neuroimage.2017.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib87] Vaz AP, Inati SK, Brunel N, Zaghloul KA. Coupled ripple oscillations between the medial temporal lobe and neocortex retrieve human memory. Science. 2019;363:975–978. doi: 10.1126/science.aau8956. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib88] Vinogradova OS. Hippocampus as comparator: role of the two input and two output systems of the hippocampus in selection and registration of information. Hippocampus. 2001;11:578–598. doi: 10.1002/hipo.1073. [DOI] [PubMed] [Google Scholar]

[bib89] Voss JL, Warren DE, Gonsalves BD, Federmeier KD, Tranel D, Cohen NJ. Spontaneous revisitation during visual exploration as a link among strategic behavior, learning, and the hippocampus. PNAS. 2011;108:E402–E409. doi: 10.1073/pnas.1100225108. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib90] Voss JL, Bridge DJ, Cohen NJ, Walker JA. A closer look at the hippocampus and memory. Trends in Cognitive Sciences. 2017;21:577–588. doi: 10.1016/j.tics.2017.05.008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib91] Watrous AJ, Lee DJ, Izadi A, Gurkoff GG, Shahlaie K, Ekstrom AD. A comparative study of human and rat hippocampal low-frequency oscillations during spatial navigation. Hippocampus. 2013a;23:656–661. doi: 10.1002/hipo.22124. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib92] Watrous AJ, Tandon N, Conner CR, Pieters T, Ekstrom AD. Frequency-specific network connectivity increases underlie accurate spatiotemporal memory retrieval. Nature Neuroscience. 2013b;16:349–356. doi: 10.1038/nn.3315. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib93] Wilming N, Onat S, Ossandón JP, Açık A, Kietzmann TC, Kaspar K, Gameiro RR, Vormberg A, König P. An extensive dataset of eye movements during viewing of complex images. Scientific Data. 2017;4:160126. doi: 10.1038/sdata.2016.126. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib94] Winters BD, Tucci MC, Jacklin DL, Reid JM, Newsome J. On the dynamic nature of the engram: evidence for circuit-level reorganization of object memory traces following reactivation. Journal of Neuroscience. 2011;31:17719–17728. doi: 10.1523/JNEUROSCI.2968-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib95] Zhang W, Sun J, Tang X. Cat head detection - How to effectively exploit shape and texture features. European Conference on Computer Vision; Marseille, France. 2008. pp. 802–816. [DOI] [Google Scholar]

PERMALINK

Hippocampal theta coordinates memory processing during visual exploration

James E Kragel

Stephen VanHaerents

Jessica W Templer

Stephan Schuele

Joshua M Rosenow

Aneesha S Nilakantan

Donna J Bridge

Roles

Abstract

Introduction

Results

Direct brain recordings linked to memory-guided eye movements

Figure 1. Direct brain recordings during memory-driven eye movements.

Table 1. Task-related eye movement behavior.

Theta dependence of memory-guided eye movements

Figure 2. Phase-locking of memory-dependent eye movements to hippocampal theta.

Figure 2—figure supplement 1. Theta oscillations at individual hippocampal electrodes.

Figure 2—figure supplement 2. Comparison of inter-trial phase coherence (ITC) between fixations to repeated and original object-locations.

Figure 2—figure supplement 3. Eye-movement related changes in theta power.

Table 2. Proportion of electrodes showing theta (4 to 6 Hz) oscillations.

Theta phase coherence increases during retrieval and novelty detection

Theta phase angle differentiates retrieval and novelty detection

Figure 3. Distinct phases of theta phase are associated with retrieval and associative novelty.

Theta phase consistency during updated-location viewing predicts subsequent memory

Figure 4. Phase-locking of hippocampal theta predicts subsequent memory.

Figure 4—figure supplement 1. Impact of saccades in pre- and post-fixation windows on inter-trial phase clustering.

Figure 4—figure supplement 2. Theta phase resets follow fixations to objects and predict subsequent memory.

Gaze-dependent theta modulations are consistent across subjects and electrodes

Table 3. Proportion of electrodes showing significant theta phase-locking effects.

Rapid sequential fixations do not account for memory-specific theta effects

Reset of hippocampal oscillations during memory-guided eye movements is specific to theta

Theta to gamma phase amplitude coupling predicts memory updating

Figure 5. Representative theta to gamma phase amplitude coupling at an individual electrode.

Figure 6. Hippocampal phase amplitude coupling predicts novelty detection and memory updating.

Memory-related changes in PAC are unrelated to theta waveform properties

Table 4. Proportion of electrodes showing changes in PAC due to theta waveform properties while viewing all objects during refresh.

Table 5. Proportion of electrodes showing changes in PAC due to theta waveform properties while viewing Mismatch objects.

Discussion

Materials and methods

Participants

Table 6. Subject demographics.

Experimental paradigm

Eye tracking

Intracranial recordings

Anatomical localization

Spectral decomposition

Phase-locking analyses

Phase amplitude coupling analyses

Statistical analyses

Acknowledgements

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Ethics

Additional files

Data availability

References

Decision letter

Roles

Author response

Author response image 1. Mean event related potentials (ERPs) across all patients and electrodes.

Associated Data

Data Citations

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases