Skip to main content
NCBI home page
HHS Author Manuscripts logoLink to HHS Author Manuscripts
. Author manuscript; available in PMC: 2021 Mar 2.
Published in final edited form as: J Clin Exp Neuropsychol. 2019 Sep 1;42(1):14–27. doi: 10.1080/13803395.2019.1659755

Spatial relational memory in individuals with traumatic brain injury

Arianna Rigon a,b, Hillary Schwarb c, Nathaniel Klooster d, Neal J Cohen c, Melissa C Duff a
PMCID: PMC7924537  NIHMSID: NIHMS1565208  PMID: 31475607

Abstract

Introduction:

Relational memory is the ability to bind arbitrary relations between elements of experience into durable representations and the flexible expression of these representations. It is well known that individuals with traumatic brain injury (TBI) have declarative memory impairments, but less is known about how TBI affects relational memory binding, the deficit at the heart of declarative, or relational, memory impairment. The aim of the current study is to examine such deficits.

Method:

We used a spatial reconstruction task (SRT) with 29 individuals with TBI and 23 normal comparison (NC) participants to investigate four different types of spatial relations: (A) identity-location relations, i.e., the relationship between a specific item and its known location; (B) item–item relations, or the relationship between one item and another; (C) item–display relations, or the relationship between an item and its position in the display; and (D) compound–item relations, i.e., relations that involve combinations of A, B, and C.

Results:

Our data revealed that individuals with TBI showed impairments in learning identity–location relations and increased compound errors compared to NCs. We also found evidence that when item identity is disregarded, individuals with TBI do not perform differently from NCs. An exploratory analysis revealed that while relational memory performance was significantly correlated with scores on the California Verbal Learning Test (CVLT), more participants with TBI exhibited impairment on the SRT than of the CVLT.

Conclusions:

Our findings show that relational memory is impaired following TBI, and provide preliminary evidence for an easy-to-administer task with increased sensitivity to memory impairment.

Keywords: Spatial memory, relational memory, traumatic brain injury, declarative memory, hippocampus

Introduction

Memory deficits are a frequent consequence of moderate-severe traumatic brain injury (TBI) and are among the most commonly recognized and treated deficits following TBI (Murray, Ramage, & Hopper, 2001; Vakil, 2005). In particular, deficits in long-term declarative memory (i.e., the encoding, storage, and retrieval of factual information and autobiographical experiences) are often reported following TBI, and these deficits interfere with skills that are necessary during the recovery stage of an injury, such as remembering to take medications, encoding the associations between names and faces of doctors and caregivers, or learning new compensatory strategies (Palacios et al., 2013; Skidmore, 2015; Tulving & Schacter, 1990). To date, in the TBI literature, declarative memory is most frequently assessed and characterized with classic neuropsychological tests of memory and learning (e.g., California Verbal Learning Test) (Palacios et al., 2013; Rigon, Turkstra, Mutlu, & Duff, 2016; Serra-Grabulosa et al., 2005).

The high prevalence of declarative memory impairment in individuals with TBI can be linked to the high vulnerability of the hippocampus and surrounding medial temporal lobe (MTL) structures, as well as the connectivity within the MTL and with the rest of the brain. For example, hippocampal damage can result from internal forces (e.g., proximity of the middle cranial fossa (Gean, 1994; Tate & Bigler, 2000) as well as from hypoxia (Atkins, 2011) and seizures (Vespa et al., 2010), both of which disproportionally affect the hippocampus relative to other brain regions. TBI is associated with hippocampal cell death, which leads to memory impairments often recorded both in experimental and clinical settings (Ariza et al., 2006; Brezova et al., 2014; Royo et al., 2006; Wang, Gao, Michalski, Zhao, & Chen, 2016). The critical role of the hippocampus and MTL in the formation of new enduring long-term declarative memories and their subsequent retrieval is well established (Eichenbaum & Cohen, 2001; Squire, 1992), both in healthy and neurological populations (Eichenbaum, 2001; Graham & Hodges, 1997; Heckers, 2001; Schuff et al., 2009). Moreover, diffuse axonal injury (i.e., widespread damage to white matter tracts) is the hallmark of TBI and can disrupt connections within and between the hippocampus and other neural systems (e.g., prefrontal cortex), as well as between different cortical systems that are necessary to encode and retrieve declarative memories (Palacios et al., 2013; Sharp, Scott, & Leech, 2014; Smith et al., 1997; Tate & Bigler, 2000).

Advances in cognitive neuroscience research over the last 20–30 years have emphasized the relational nature of declarative memory and demonstrated that at the heart of the declarative, or relational, memory deficit following hippocampal damage is an impairment in relational memory binding. Relational memory theory emphasizes two hallmark features of the hippocampus in relational memory (Eichenbaum & Cohen, 2001; Rubin, Schwarb, Lucas, Dulas, & Cohen, 2017): (1) the binding of arbitrary relations between the elements of experience into durable representations of events or scenes; and (2) the flexible expression of these representational bindings in novel settings, i.e., those contexts different from the original encoding context; this allows for the search, reconstruction, and recombination of the information contained within them (as opposed to a “video-camera”-like recapitulation of prior events) and for use in novel situations (Konkel & Cohen, 2009). Thus, the compositional nature of relational memory representations allows for the retrieval or reactivation of individual, or even multiple configurations of associations that exist among the rich and complex experiences of our daily lives and for the flexible use of this knowledge across a broad range of conditions (Eichenbaum & Cohen, 2014). For instance, relational memory representations permit us to learn the names of new acquaintances (completely arbitrary relations that cannot be appreciated a priori or inferred upon the initial encounter), to remember where and at what time to go for a new appointment (relying on representations previously created from past learning experiences and flexibly using them prospectively in service of a new goal), and even to facilitate new learning of information (Rubin et al., 2017). Attention to the flexible expression of relational memory, together with the development of novel paradigms, methods, and measures to assess relational memory representations has also revealed the critical role relational memory plays in the dynamic and adaptive use of memory and in flexible cognition and goal-directed behavior more broadly (Rubin, Watson, Duff, & Cohen, 2014). Indeed, building on a large body of work on the nature spatial memory in healthy individuals and spatial memory deficits following different forms of neurological injury (Kessels, de Haan, Kappelle, & Postma, 2002; Postma, Kessels, & van Asselen, 2008; Postma, Morel, Slot, Oudman, & Kessels, 2018), our group has been actively involved in the development of new experimental tasks that directly test the relational binding and flexibility of hippocampal-dependent representations permitting the detection of relational memory deficits that may not be routinely captured with traditional, standardized assessments of declarative memory (i.e., those that emphasize encoding and subsequent retrieval via recalling verbatim word lists or producing exact replication of figures).

Considering that memory is crucial to the development of compensatory skills and to the re-learning of abilities necessary for independence and community reintegration (e.g., through instructional learning and metacognitive approaches; for a review, see Skidmore, 2015), a better understanding of the specific patterns and profiles of memory deficits that follow a TBI can help shed light on multi-domain impairments, and direct new rehabilitative efforts.

Here, we used a spatial reconstruction task (SRT) to examine relational memory, one of our experimental tasks designed to capture relational memory (although it should be noted that this task places higher demands on relational memory binding and encoding than on the flexible express of such representational bindings in novel settings). While it is likely that spatial relational memory is not the only type of relational memory impaired in TBI, we chose this task because of its ease of administration, and because it has been previously employed to examine relational memory in populations with neurological damage (Horecka et al., 2018). In SRTs, participants study displays with a varying number of items presented at different spatial locations (Horecka et al., 2018; Schwarb et al., 2017; Watson, Voss, Warren, Tranel, & Cohen, 2013). The items then disappear from view and after a short delay the items reappear aligned at the top of the display and participants are asked to reconstruct the original con-figuration by freely placing each item in the original location (See Figure 1). A growing body of work using SRTs has linked performance with hippocampal integrity and memory ability in healthy and disordered populations (Clark et al., 2017; Horecka et al., 2018; Watson et al., 2013). The SRT is particularly powerful in that it provides multiple dependent measures that dissociate different types of relational information that are present in each trial.

Figure 1.

Figure 1.

Open in a new tab

The spatial reconstruction task.

Recently, Horecka et al. (2018) established an analysis pipeline that identifies multiple, dissociable measures of interest that each contribute to successful memory performance. There are three classes of measures: (1) Global Reconstruction Errors, (2) Misplacement Errors, and (3) Identity–Location Relations. Global Reconstruction Errors are a class of measures includes systematic errors that affect the overall reconstruction. Three global error measures are considered: (a) Rotation (spinning of the entire display), (b) Scaling (squishing and stretching of the entire display), and (c) translation (x-y shifting the entire display). Misplacement Errors quantify the distance an item was placed from its studied position. Misplacement errors can be calculated on the reconstructed display, but also after correction for each of the global errors independently or in a stepwise fashion. Once global errors have been accounted for and corrected, several measures of specific relational bindings can be computed. These Identity–Location Relations are a class of measures that track local errors that occur between one or more objects in the display. They include (a) identity–location relations, or the relationship between a specific item and it’s known location (i.e., knowing that that item A belongs in location A); (b) item–item relations, or the relationship between one item and another item (i.e., knowing that item A belongs to the right of item B); (C) item–display relations, or the relationship between an item and its general position in the display (i.e., knowing that item A belongs in the upper right corner of the screen); and (D) compound–item relations, or the relationship between item–identity relations and item–item relations (i.e., knowing that item A belongs in location A, item B belongs in location B, and item A belongs to the right of item B). While related, each of the Identity–Location Relations highlight a different relationship between the items and each other and the items and the spatial arrangement of those items. In the current study, we examined how relational memory, and in particular spatial relational memory, is impacted in individuals with moderate-severe TBI using the same methods and measures used by Horecka et al. (2018) in individuals with hippocampal amnesia.

A recent study considered these outcome measures comparing a group of patients with hippocampal damage and profound memory impairment (hippocampal amnesia) to a group of matched, healthy comparison individuals (Horecka et al., 2018). In that study, the hippocampal patients were not impaired on all outcome measures, but rather showed a specific pattern of deficits: amnesia predominantly leads to deficits in binding the arbitrary relationship between a specific item and its location (i.e., identity–location relations). Conversely, individuals with hippocampal amnesia were able to represent the relationships between items independent of their identity (item–item relations) and the global position of various items (item–display relations). These data suggest that even in the context of a profound relational (declarative) memory impairment, there are specific patterns of spatial errors that emerge.

Although TBI routinely causes hippocampal damage and memory impairments, TBI seldom causes the same profound and focal deficit in memory, or amnesia, as that found in patients who suffered from prolonged anoxic events or herpes simplex encephalitis (such as the patients examined in Horecka et al., 2018). Moreover, other types of neural damage occur after TBI, such as diffuse axonal injury and white matter damage, as well as atrophy in multiple cortical and subcortical brain regions (Bigler, 2013), which could lead to different profiles of impairment in spatial relational memory in TBI than has been reported in patients with hippocampal damage. Here, we investigated performance on the SRT in a group of individuals with moderate-severe TBI, to examine their ability to bind together and recall various types of relational information (e.g., item–item relations, item–display relations, identity–location relations, as well as compound relations). The aim of the current work was to investigate the presence of specific patterns of spatial memory errors, thus shedding light on the type(s) of spatial relational memory impairment that may be common in TBI and that could possibly be targeted by rehabilitation efforts following TBI. The current work also expands our knowledge of hippocampal-dependent memory functions following TBI. Given the longstanding literature on hippocampal pathology in TBI and that performance on the SRT is highly correlated with hippocampal integrity, we might expect a similar pattern of deficits as found in patients with amnesia, although perhaps less severe. On the other hand, considering the diffuse nature of TBI, and the numerous neural and cognitive systems that can be affected, TBI may produce a unique pattern of deficit above and beyond those observed as a result of impaired relational memory and hippocampal pathology. Additionally, this work includes an exploratory analysis to examine the relationship between performance on the SRT and other neuropsychological assessments of memory and learning within the TBI group; in particular, we administered tests of set shifting, processing speed, and memory, and examined their correlations with SRT measures within the TBI group; we also investigated whether the SRT has increased sensitivity to memory impairment than traditional tests of declarative memory.

Methods

Participants

Participants were 29 individuals with moderate-severe TBI and 23 normal healthy comparison (NC) participants. Participants were recruited through the University of Iowa community and through the University of Iowa Brain Injury Registry (Rigon et al., 2016; Rigon, Voss, Turkstra, Mutlu, & Duff, 2016, 2017). Inclusionary criteria for individuals with TBI were (1) history of moderate to severe TBI, (2) chronic post-injury phase (all participants were >6 months post injury). Language and motor deficits that would have prevented participants from correctly understanding instruction or participating in the study were ruled out. The TBI and NC groups did not significantly differ for age (t(50) = 1.03, p> .05), education t(50) = −1.5, p> .05), or sex (X2(1, N = 52) = 2.4, p> .05) (Table 1).

Table 1.

Participant demographics.

N AGE (Mean±SD) SEX (Females) EDUCATION (Mean±SD) CHRONICITY (Months, Median±IR)
NC 23 53.96 ± 14.51 12 15.13 ± 1.74 N/A
TBI 29 49.84 ± 14.01 9 14.24 ± 2.29 31.5(±38)
Group Differences (p) N/A .31 .12 .13 N/A
Open in a new tab

NC = Healthy comparison participants, TBI = Traumatic brain injury, p = p-value, SD = Standard Deviation, N/A = Not Applicable, IR = interquartile range

Time since injury spanned from a minimum of 6 months and a maximum of 307 months before testing. One participant had sustained two separate TBIs. Causes of injury were falls (14), motor (10) and non-motor vehicle accident (2) and hits/assaults (4).

We assessed TBI severity using the Mayo Classification System (Malec et al., 2007). Participants were considered having sustained a moderate-severe injury if at least one of the following criteria was met: (1) Glasgow Coma Scale (GCS)<13 (i.e., moderate or severe according to the GCS), (2) positive acute CT findings or lesions visible on a chronic MRI, (3) loss of consciousness (LOC)>30 min or post-traumatic amnesia (PTA)>24 h. Injury-related information was collected using available medical records as well as a semi-structured interview with the participants. GCS was available for 14 participants (Median = 13, ranging from 3 to 15); LOC information were available for 16 participants; PTA was available for 15 participants; acute imaging information was available for 27 participants. Cause of injury were falls (14), motor vehicle accidents (12), being hit or assaulted (4), or non-motor vehicle accidents (2).

NCs were included if they had no self-reported history of head injury or loss of consciousness or no history of neurological, psychiatric or developmental disorders. Informed consent was obtained for experimentation with human subjects.

Spatial reconstruction task

Patients completed a computerized SR task (Monti et al., 2015; Schwarb, Johnson, McGarry, & Cohen, 2016). On each trial, participants studied a display in which five abstract items appeared at unique locations on the computer screen. Abstract items were selected to limit verbal coding of the items. The display was studied for 20 s before the items disappeared for 4 s. The items then reappeared in a line at the top of the screen, and participants used a mouse to move each item back to the position in which it was studied. Note that when objects re-appear on the top of the screen the order is randomized. Due to randomization, it is possible that item–item relations may or may not be disrupted, but this is not systematic in this experiment. Participants were permitted to make as many adjustments to their reconstruction as desired and hit the spacebar to indicate that they were satisfied with the accuracy of their reconstruction. There were 15 trials (See Figure 1).

Statistical analyses: Dependent variables

Following Horecka et al. (2018), data analysis involved a comparison of the original studied display and the participant’s reconstructed display. The analysis entailed four steps: (1) Identity remapping: The identity of each stimulus was stripped and data were reduced to a set of coordinates; reconstruction coordinates were then compared to the study display coordinates; (2) Global error correction: Systematic spatial errors that were shared across all items in the display (i.e., global rotation, global scaling, and global translation) were computed and corrected so that relational information was directly comparable between the studied display and the reconstructed display; (3) Location placement evaluation: Identity-stripped, global error-correct item placement was compared to identity-stripped studied item location based on an accuracy threshold (95% confidence intervals for the distribution) to determine if any item was placed in a correct studied location; and finally (4) Compound error evaluation: The identity of items is reintroduced to determine whether two items were assigned each other’s location (swaps) or several items were assigned in locations that were correct, but that were originally studied for other items (cycles). While swaps are technically the simplest form of cycles, we considered the two measures separately for consistency with previous work (Horecka et al., 2018). The analysis pipeline here described yielded the dependent variables discussed below (See Figure 2 and Table 2)

Figure 2.

Figure 2.

Open in a new tab

The analysis pipeline for the spatial reconstruction task.

Table 2.

Dependent variables on the Spatial Reconstruction Task.

Measure Type
Overall misplacement Local misplacement
Post-remapped misplacement
Post-transformed misplacement
Translation Global error
Scaling
Rotation
Accurate single item placements Identity-location relations
Accurate misassignments
Inaccurate location Item-item and item-display relations
Swaps, Cycles Identity-location, item-item and item-display relations
Open in a new tab

Measures of local misplacement

First, we examined Overall misplacement, a global measure obtained by calculating the sum of the Euclidean distance between each item’s original studied location and the location that item was placed in by the participant. Both older adults and individuals with hippocampal amnesia tend to show higher levels of misplacements than younger or healthy adults (Clark et al., 2017; Horecka et al., 2018; Monti et al., 2015). This indicates that misplacement is a measure that is sensitive to hippocampal function, although it does not reveal information about the specific nature of deficits in spatial relational memory. Higher values correspond to higher degrees of misplacement. While misplacement has been used as a measure of relational memory in the past (Watson et al., 2013), misplacement values can be inflated by errors that are not specifically related to relational memory (i.e., to how items relate to each other or to their location). For instance, misplacement values are higher when participants move all items up or down when reconstructing, or by rotating them. In the current study, in addition to overall misplacement error, we examined Post-remapped misplacement, which is calculated after identity remapping (i.e., step (1) above), and Post-transformed misplacement, which is calculated after global error correction (i.e., step (2) above). Misplacement error variables were measured in pixels. Remapped error (i.e., the difference between overall misplacement and post-remapped misplacement) and transformed error (i.e., the difference between post-remapped and post-transformed misplacement) were also computed for individuals with TBI and NCs separately. These error measures reveal the amount of error accounted for by identity-remapping and global error correction.

Measures of global error

We also examine, separately, the transformation components that account for the global error correction in step (2) of the analysis: translation (i.e., the offset of an item’s placement compared to its studied position that can be interpreted as a misrepresentation of the stimulus space in the overall display), which is measured in pixels; scaling (extending or shortening of the stimulus space in the display), measured as a proportion where values <1 indicating shrinking and values >1 indicating stretching; and rotation (a misrepresentation of the directions of the stimulus space in the display), which is measured in radians.

Measures of identity–location relations

We examined Accurate single item placements, a measure of the ability to place the correct item in the correct location (i.e., the binding between an item and its specific location, as opposed to another item); this is the strictest accuracy measure as it requires that the correct item be placed in the correct location. We also examined Accurate misassignments, which is an index of how many items were placed in the studied location that originally corresponded to another item (i.e., a correct item-location but a wrong identity-location); this is a measure of the ability to remember accurate location regardless of identity. Note that these measures were calculated after compound error evaluation (i.e., step (4) described above).

Measure of item-item and item–display relations

Inaccurate location is a measure that summarizes all items that were not placed in any studied location (i.e., neither their original location nor a location that was originally another item’s). This measure is corrected for global error. Again, these measures were calculated after compound error evaluation.

Measures of compound errors (identity-location, item-item, and item–display relations)

Finally, we specifically examined the presence of Swaps (pairs of items that were inverted) and Cycles (groups of more than two items that were assigned to each other’s correct locations). These measures were calculated after compound error evaluation.

Neuropsychological testing

Participants were administered (1) the California Verbal Learning Task (CVLT) (Delis, Freeland, Kramer, & Kaplan, 1988), to assess immediate recall (CVLT-IR), short-delay verbal recall (CVLT-SR) and long-delay verbal recall (CVLT-LR); (2) the Trail Making Test – Part B (TRAIL B) (Gordon, 1972) to obtain normalized scores of executive functioning (Tombaugh, 2004); and (3) the Coding and Symbol Search subtests of the Wechsler Adult Intelligence scale IV (Holdnack, Xiaobin, Larrabee, Millis, & Salthouse, 2011) to obtain a processing speed index (PSI). Neuropsychological scores were only available for TBI participants (because of their participation in the Iowa TBI Registry) and some measures were missing for a subset of the participants: CVLT (missing for four individuals with TBI), Trails (missing for four individuals with TBI), and PSI (missing for three individuals with TBI). For two participants with TBI, the Trails B was discontinued, and a Windsorization approach was used so that their Trails B scores were recoded into the lowest z-scores found in the overall dataset (Costa, 2014) (Table 2).

Statistical analysis

Measures of local misplacement (Overall misplacement, Post-remapped misplacement, and Post-transformed misplacement) were analyzed using a repeated measure Analysis of Variance (ANOVA), with Group as a between factor and Misplacement Type as a within factor. Two-tailed independent and paired-sample t-tests were used for follow-up analyses. Bonferroni correction for multiple comparison was applied for post-hocs. We opted for a repeated measure ANOVA for consistency with Horecka et al. (2018) but it should be noted that it is possible that Error measures are, in fact, derived from each other (the results of the independent and paired-sample t-tests should then be considered). The same ANOVA was used to examine group differences in Misplacement error, with Error Type as the within factor and Group and the between factor. For the three measures of global error (Translation, Scaling, and Rotation), two-tailed independent and paired-sample t-tests were used.

For Measures of identity–location relations (Accurate single-item placements, Accurate misassignments) as well as Inaccurate locations and measures of compound errors (Swaps and Cycles), two-tailed independent sample t-tests were used, applying a Bonferroni correction for multiple comparisons.

We also examined, both within the TBI and the NC group, the presence of significant Pearson’s correlations between some dependent variables yielded by the SRT and the demographic variables (age and education) as well as with neuropsychological variables for the participants with TBI. The dependent variables we focused on were Post-transformed misplacement (as it is a measure of misplacement correct for error shared across all items); Accurate single item placement (as it can be considered a measure of identity-location memory success); and Inaccurate location, which is a measure of both item-item and item-display performance. The measures we included in the correlational analyses were chosen because they are thought to best represent relational memory success and error, as well as memory for location regardless of identity.

Lastly, we carried out an exploratory observational analysis to compare performance on a widely used test of declarative memory (the CVLT) and on three SRT measures Post-transformed misplacement, Accurate single item placement, and Inaccurate location. All measures were normally distributed (all ps>.124). In particular, we examined whether SRT indices might be more sensitive in detecting deficits in relational memory ability than the CVLT by comparing the number of individuals with TBI who were impaired in the task (using the liberal threshold of performance ≤1.5 standard deviation below average). For the CVLT, T and Z scores where used. For the SRT measures, scores were standardized using the mean and standard deviation of the NC group. The number of participants who performed 1.5 standard deviations below 0 (or, in the case of Post-transformed misplacement and Inaccurate location, above, as they are error measures) was then counted.

Results

Measures of local misplacement

A repeated measure ANOVA showed a significant effect of Group (F1,50 = 11.89, p = .001, ηp2 = .19), with individuals with TBI showing misplacements of higher magnitude at every Misplacement Type (overall misplacements, post-remapped misplacements, post-transformed misplacements); a significant effect of Misplacement Type (F2,100 = 133.63, p < .001, ηp2 = .73), with misplacement magnitude decreasing with each additional correction; as well as a significant Group by Misplacement Type interaction (F1.04,100 = 5.74, p = .02, ηp2 = .1). Individuals with TBI showed higher magnitude of misplacement at all stages (All T(50)<−3.1, all p < .003, all d > .87). There was also a significant decrease in misplacement for both groups after both steps of the analyses (For NC, all T(22)>5.4, all p < .001; for TBI, all T(28)>9.08, all p < .001). However, further examination of the interaction, via visual inspection of the data, shows that the group difference decreased steeply post remapping (i.e., after stripping items of their identity), and remained lower post transformation (i.e., after accounting for global error such as rotation, scaling and translation) (see Figure 3).

Figure 3.

Figure 3.

Open in a new tab

Group differences in misplacement error across the three steps of the analysis (i.e., remapped errors, following identity stripping, and transformed error, which accounts for rotation, scaling and translation). Individuals with TBI showed error of larger magnitude across all steps.

We also examined group differences in misplacement errors (i.e., remapped errors, following identity stripping, and transformed error, which accounts for rotation, scaling and translation). We found a significant effect of Group (F1,50 = 6.63, p = .01, ηp2 = .12) with TBI patients showing error of larger magnitude than NCs and of Type of Error (F1,50 = 67.03, p < .001, ηp2 = .57) with Remapped errors being larger than transformed errors, but no significant Group by Type of Error interaction (F1,50 = 2.84, p = .1, ηp2 = .06).

When we examined the three measures of global error (translation, rotation, and scaling) separately, a t-test showed a significant group difference in translation, revealing that individuals with TBI performed translations of higher magnitude than NCs (T(50) = −3.49, p = .001, d = 1.01). There was no group difference in scaling, indicating that individuals with TBI did not distort (i.e., stretch or squish) items in their stimulus space more than NCs (T(50) = .93, p = .36, d = .26). However, there was a significant group difference in rotation, indicating that individuals with TBI tended to rotate the display more than NCs (T(50) = −3.75, p = .008, d = .8) (see Figure 4).

Figure 4.

Figure 4.

Open in a new tab

Group differences on the three measures of global error (translation, rotation, and scaling). There was a significant group difference in translation and rotation, but not in scaling.

Measures of item–identity relations

We found a significant group difference on Accurate single item placements (T(50) = −2.94, p = .005, d = .82), where individuals with TBI placed the correct item in the studied location less frequently than NCs. For Accurate misassignments, too, there was a significant group difference (T(50) = −3.08, p = .003, d = .86). In this case, individuals with TBI more frequently placed items in a location that was originally the location of another item. When we examined Inaccurate location, we found no significant group difference (T(50) = .38, p = .71, d = .11). This indicates that individuals with TBI did not place items in a studied location (regardless of whether it was that specific item or another item’s location) more or less frequently than healthy comparison participants (see Figure 5).

Figure 5.

Figure 5.

Open in a new tab

Group differences on Accurate single item placement, Accurate misassignment and Inaccurate location. Individuals with TBI produced significantly fewer Accurate single item placement and more Accurate misassignment, but there was no group difference on Inaccurate location.

Higher level compound relations

Individuals with TBI produced significantly more Swaps than NCs (T(50) = −2.5, p = .02, d = .17). However, they did not produce significantly more Cycles (T(50) = −1.95, p = .06, d = .4). This indicates that while individuals with TBI were more likely to invert two items, they were no more likely to interchange the position of three or more items. It should be noted, however, that there the latter statistical test revealed a trend toward significance (p = .06) (see Figure 6).

Figure 6.

Figure 6.

Open in a new tab

Group differences in Swaps and Cycles. There was a group difference (i.e., individuals with TBI produced significantly more) in Swaps, but not in Cycles.

Correlational analysis and comparison with neuropsychological tasks

Within the NC group, when correcting for multiple comparisons the only significant correlation was between age and Post-transformed misplacement (r = .7, p < .001), with older individuals showing higher levels of misplacements. Within the TBI group, there were no significant correlations between any of the demographic or injury-related variables and the three spatial memory indices. A Fisher r-to-z transformation showed a significant difference in correlation coefficients measuring the relationship between age and Post-transformed misplacement was different between TBI (r = .31, p > .05) and NCs (z = .02, p = .02).

All neuropsychological measures were significantly correlated with both Post-transformed misplacement (negative correlation) and with Accurate single item placement (positive correlation). However, there was no significant relationship between Inaccurate location and any of the neuropsychological measures. (See Table 3 & 4)

Table 3.

Performance on neuropsychological tasks in the TBI group.

Mean, SD N
CVLT-IR 50.8, 13.12 25
CVLT-SR .0, 1.16 25
CVLT-LR .0, 1.26 25
TRAIL B −1.59, 1.62 25
PSI 95.27, 12.49 26
Open in a new tab

CVLT-IR = California Verbal Learning Task, immediate recall (T scores), CVLT-SR = California Verbal Learning Task, short recall (Z scores), CVLT-LR = California Verbal Learning Task, long recall (Z scores), TRAIL B = Trail Making Test – Part B, PSI = Processing Speed, N = number of participants for whom the neuropsychological assessment was available.

Table 4.

Correlation between SRP measures and neuropsychological tests in the TBI group.

Post-transformed Misplacement Accurate single item placement Inaccurate location N
CVLT-IR −.64, .001 .67, <.001 −.03, .91 25
CVLT-SR −.6, .001 .6, .002 −.01, .96 25
CVLT-LR −.56, .003 .6, .002 −.01, .96 25
TRAIL B −.56, .003 .56, .003 −.08, .72 25
PSI −.58, .002 .55, .004 −.01, .9 26
Open in a new tab

CVLT-IR = California Verbal Learning Task, immediate recall, CVLT-SR = California Verbal Learning Task, short recall, CVLT-LR = California Verbal Learning Task, long recall, TRAIL B = Trail Making Test – Part B, PSI = Processing Speed, N = number of participants for whom the neuropsychological assessment was available. The table reports Pearson’s r and p-value (r, p).

Lastly, our exploratory analysis aimed at comparing performance on the CVLT and the SRT, revealed that the number of participants who performed at least 1.5 standard deviations below average was 3 for the CVLT-IR, 3 for the CVLT-SR, and 4 for the CVLT-LR. Note that the participants who performed 1.5 standard deviations below the mean were the same for all three measures of the CVLT, with one additional participants showing impairment in on the CVLT-LR.

For Post-transformed misplacement, eight participants performed 1.5 standard deviations above average (i.e., made more errors). The number of participants dropped to six when only participants for whom the CVLT was available were considered. All participants who were impaired on the CVLT-IR and the CVLT-SR were also impaired on Post-transformed misplacement. Notably, three additional participants with TBI were impaired on this measure that were not identified by the CVLT.

For Accurate single item placement, nine participants performed 1.5 standard deviations below average (i.e., made fewer errors). The number of participants dropped to seven when only participants for whom the CVLT was available were considered. All participants who were impaired on the CVLT-IR and the CVLT-SR were also impaired on Post-transformed misplacement. Again, four additional participants with TBI were impaired on this measure that were not identified by the CVLT.

For Inaccurate Location, five participants performed 1.5 standard deviations above average (i.e., more inaccurate locations). The number of participants dropped to four when only participants for whom the CVLT was available were considered. Here, there was no overlap between participants who were impaired on the CVLT-IR and the CVLT-SR, but the one participant who was impaired on the CVLT-LR was also impaired on Inaccurate Location.

Discussion

We investigated how moderate-severe TBI impacts different aspects of spatial relational memory: identity-location, item-display, and item-item, as well as misplacement. Individuals with TBI showed impairment specifically in the binding of identity–location relations. They also show larger local misplacement than NCs and more compound errors (i.e., errors that combine the three different types of relations). In addition, an exploratory analysis revealed that performance on the SRT, and in particular Misplacement and Accurate single item placements (i.e., a measure of identity-location memory success) correlated with scores on standard neuropsychological assessments of memory and learning; however, more individuals were identified as memory impaired on the SRT than on the CVLT. We discuss each specific finding in depth below.

Horecka et al. (2018) found that patients with hippocampal amnesia show preserved binding of item-display and item–item relations in the SRT: this was measured by examining the number of inaccurate locations (i.e., items placed in locations that were wrong, independent of their identity). In the present study, we found that when identity information is ignored the number of inaccurate locations does not differ between individuals with TBI and NCs. This reveals that a gestalt-like representation of the items (such as the overall shape formed by all the items) may be maintained in individuals with TBI. This shape-like representation does not rely on identity-location information (because item identity is not taken into account when determining whether placement is correct or not) but exclusively on a combination of item-item and item-environment display relations. These two types of spatial relational information contain redundant elements that give participants the ability to place an item (independent of its identity) in the correct location; this is especially true with the analysis pipeline used here, that corrects for individual noise, or global error, and that attributes a binary value to participants’ response based on an accuracy threshold (i.e., correct vs. non-correct), thus stripping the data of potential confounds. This means that when reconstructing the global shape of the display originally studied, individuals with TBI have at their disposal multiple pieces of relational information (item-item and item-display information). However, when individuals with TBI have to rely on entirely arbitrary identity–location relations to reconstruct the display, they are impaired compared to NCs and place fewer items in their correct locations and more items in a location that belonged to another item during study. This pattern (higher Accurate misassignments, lower Accurate single item placements compared to NCs) is similar to the pattern observed in patients with hippocampal damage. Indeed, identity-location information has been identified as the most hippocampal-dependent information in SR tasks (Horecka et al., 2018), and this is in line with relational memory theory and its view of the hippocampus as necessary for fast binding of arbitrary elements.

Individuals with TBI also showed significantly more misplacement errors than NCs at every step of the analysis: both when stripping the data of identity information, and when correcting for global error. The largest decrease in misplacement was found after step 1 of the analyses (i.e., after stripping identity information), but the magnitude of misplacement further decreased for both groups after global error transformation. What is interesting is that not only the Remapped Error but also the Transformed Error were significantly different between groups. This is different from what has been reported with patients with hippocampal damage, who showed significantly more Remapped Error, but the same rates of Transformed Error, than NCs (Horecka et al., 2018). Further examination of global error revealed that while there was no significant difference in scaling, individuals with TBI presented rotation errors of higher magnitude, as well as more translation errors. These types of global errors are important to consider, because they are shared across all items, and they are not a specific indication of spatial relational memory or hippocampal-dependent representation (although individuals with hippocampal damage do show more rotation errors, but whether this can be specifically ascribed to hippocampal damage is unclear). We speculate that it is possible that it’s not (or not solely) hippocampal damage that drives higher rates of these types of errors in TBI. As TBI is often accompanied by both gray matter atrophy in various cortical and subcortical regions and diffuse white matter damage outside of the MTL, there are a variety of possible neurobiological explanations for the higher rates of global error. It should also be kept in mind that even when accounting for these higher rates of share errors between the items, individuals with TBI still showed higher rates of local misplacements (post-transformed misplacement).

Another interesting observation is that individuals with TBI also tended to make significantly more swaps, which are compound spatial relational memory errors. While this is not true about cycles, it should be noted that there were large individual differences within the TBI group, and a strong trend toward significance, and that, on average, individuals with TBI did show more cycles that NCs. This is particularly notable because when the analysis pipeline used here is employed, individuals with hippocampal amnesia, who have profound declarative memory impairment, do not make more swap errors than NCs. Swap errors require the ability to bind two locations and their positions relative to each other and the display. While counterintuitive on the surface, we speculate that the increased swap errors in individuals with TBI relative to patients with hippocampal amnesia are due to a less severe deficit in relational memory. That is, individuals with TBI are able to retain enough spatial relational memory (such as a memory for two locations and their relative positions) to make these types of errors more likely than errors that are due to chance.

When we examined the correlation between different neuropsychological measures (processing speed, executive functioning, and verbal learning and memory) we found that our measure of identity-location memory, as well as our measure of overall misplacement, were respectively positively and negatively associated with all neuropsychological scores. Interestingly, Inaccurate location, which is an overall measure of error that doesn’t map onto relational memory, did not correlate with any of the CVLT scores. Notably, when we examined how many participants with TBI were impaired on the CVLT and on Accurate single item placements, we found that 100% more individuals with TBI (6 vs. 3) were impaired. While the approach used here contains limitations (for instance the fact that neuropsychological data were not available for the comparison group), these results lead us to speculate that it is possible that our experimental tasks of relational memory are more sensitive to detecting hippocampal pathology and/or memory disruption than current standardized neuropsychological tests of memory. Additional evidence in support of this comes from the fact that another test of relation memory (face-scene paradigm) detected relational memory deficits in individuals with mild TBI (Monti et al., 2015). SRTs, in particular, provide other advantages, as they are easy and quick to administer and do not require verbal responses, and are thus ideal for clinical assessment of individuals with more severe injuries.

The results here provide initial evidence for disruptions in relational memory in individuals with moderate-severe TBI and may lead to advances in understanding the underlying mechanisms of behavioral dysfunction in the everyday lives of individuals with TBI. Beyond canonical memory functions, disruptions in relational memory are also associated with inflexible cognition and maladaptive behavior in a surprisingly broad way across cognitive domains: in language and communication (Duff & Brown-Schmidt, 2012), social cognition (Beadle, Tranel, Cohen, & Duff, 2013; Davidson, Drouin, Kwan, Moscovitch, & Rosenbaum, 2012; Spreng, 2013), decision-making (Gupta et al., 2009), and navigation and active environmental exploration (Maguire, Nannery, & Spiers, 2006; Voss, Gonsalves, Federmeier, Tranel, & Cohen, 2011; Voss et al., 2011; Yee et al., 2014). The results of our exploratory analysis raise the possibility that traditional assessments (e.g., CVLT) may not be sensitive to the full range of relational memory capacities, including those required for aspects of flexible and adaptive cognition, and that the use of such assessments alone may underestimate the presence and impact of relational memory deficits in the outcomes of those with TBI. Future work should explore these possibilities by examining the relationship between performance on this task, as well as other relational memory tasks, and aspects of flexible cognition and adaptive behavior, and whether such relations have predictive value in determining long-term behavioral outcome. In particular, to be able to use measures such as the SRT within clinical settings, we will need to condense the outcome measures, and to determine which dependent variables are most likely to map onto real life memory deficits and on overall outcome following TBI. Our lab is currently working on this in a new and larger sample of individuals with moderate-severe TBI.

A limitation of this study is the lack of sufficient neuroimaging or neuroanatomical data to assess the relation between the structure and function of the hippocampus and task performance. Although the patterns of disruption in TBI were strikingly similar to those patients with focal hippocampal damage reported by Horecka et al. (2018), given the widespread nature of neural damage in TBI, it is highly likely that the participants here have damage that extended to other cortical and subcortical gray and white matter regions. The neuroanatomical complexity present in TBI underscores the need of future studies to examine the relationship between patterns on brain damage following TBI and SRT performance. Such studies will be able to clarify the role of the hippocampus and its connecting white matter tracts in relational memory impairment following TBI. We suggest that work examining the association between volume and microstructural integrity of specific hippocampal subfields and sub-regions, and performance on relational memory tasks, may be a more appropriate approach than examining hippocampal volume alone. Indeed, early studies of hippocampal involvement in TBI focused on the volume of the entire structure but found only modest associations of total hippocampal volume with performance on classic neuropsychological tests of memory (Bigler et al., 1996; Tate & Bigler, 2000). Considerable new knowledge has been gained about the hippocampus at various levels of organization, from its microstructural properties, to its subfields and sub-regions as well as the interactions of the hippocampus with cortical networks. Recent work in mild TBI demonstrates differential susceptibility to hippocampal subfields in relation to memory (Leh et al., 2017), pointing to the power of such techniques. Another limitation is the possible lack of generalizability of results, considering both the heterogeneity of TBI and the fact that it can be assumed that to a degree the study sample included individuals who were able and chose to participate in research, and thus who are on the “better recovered” end of the spectrum among individuals with moderate-severe TBI. Lastly, the lack of neuropsychological data for the entire sample, as well as the fact that constructs like working memory and attention were not assessed, prevented us from fully characterizing the sample and from a better understanding of the cognitive domains that are associated with SRT performance.

In conclusion, the current study examined how spatial relational memory is impacted by moderate-severe TBI. We found that while identity–location relations are impaired, item-item and item–display relations are preserved at least to a degree, and that compound spatial relational errors were more frequent. Future work interested in examining how memory is affected by TBI should focus on expanding these findings, and on determining the presence of sub-groups based on individual differences within TBI populations; moreover, they should investigate the relationship between spatial relational memory and other types of memory, and with neuroimaging findings that assess both hippocampal structure and the integrity of white matter pathways that connect it to other brain regions. The results of the current study indicate the identity-location spatial relations are most likely to be impacted by TBI and this deficit may have specific consequences for community reintegration and long-term outcome.

Footnotes

Disclosure statement

No potential conflict of interest was reported by the authors.

References

  1. Ariza M, Serra-Grabulosa JM, Junque C, Ramirez B, Mataro M, Poca A, … Sahuquillo J (2006). Hippocampal head atrophy after traumatic brain injury. Neuropsychologia, 44(10), 1956–1961. [DOI] [PubMed] [Google Scholar]
  2. Atkins CM (2011). Decoding hippocampal signaling deficits after traumatic brain injury. Translational Stroke Research, 2(4), 546–555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Beadle JN, Tranel D, Cohen NJ, & Duff MC (2013). Empathy in hippocampal amnesia. Frontiers in Psychology, 4, 69. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bigler ED (2013). Traumatic brain injury, neuroimaging, and neurodegeneration. Frontiers in Human Neuroscience, 7, 395. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bigler ED, Johnson SC, Anderson CV, Blatter DD, Gale SD, Russo AA, … Hopkins RO (1996). Traumatic brain injury and memory: The role of hippocampal atrophy. Neuropsychology, 10(3), 333. [Google Scholar]
  6. Brezova V, Moen KG, Skandsen T, Vik A, Brewer JB, Salvesen O, & Haberg AK (2014). Prospective long-itudinal MRI study of brain volumes and diffusion changes during the first year after moderate to severe traumatic brain injury. Neuroimage Clinical, 5, 128–140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Clark R, Tahan AC, Watson PD, Severson J, Cohen NJ, & Voss M (2017). Aging affects spatial reconstruction more than spatial pattern separation performance even after extended practice. Hippocampus, 27(6), 716–725. [DOI] [PubMed] [Google Scholar]
  8. Costa PJ (2014). Truncated outlier filtering. Journal of Biopharmaceutical Statistics, 24(5), 1115–1129. [DOI] [PubMed] [Google Scholar]
  9. Davidson PS, Drouin H, Kwan D, Moscovitch M, & Rosenbaum RS (2012). Memory as social glue: Close interpersonal relationships in amnesic patients. Frontiers in Psychology, 3, 531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Delis DC, Freeland J, Kramer JH, & Kaplan E (1988). Integrating clinical assessment with cognitive neuroscience: Construct validation of the California verbal learning test. Journal of Consulting and Clinical Psychology, 56(1), 123–130. [DOI] [PubMed] [Google Scholar]
  11. Duff MC, & Brown-Schmidt S (2012). The hippocampus and the flexible use and processing of language. Frontiers in Human Neuroscience, 6, 69. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Eichenbaum H (2001). The hippocampus and declarative memory: Cognitive mechanisms and neural codes. Behavioural Brain Research, 127(1–2), 199–207. [DOI] [PubMed] [Google Scholar]
  13. Eichenbaum H, & Cohen NJ (2001). From conditioning to conscious recollection: Memory systems of the brain. New York, NY: Oxford University Press. [Google Scholar]
  14. Eichenbaum H, & Cohen NJ (2014). Can we reconcile the declarative memory and spatial navigation views on hippocampal function? Neuron, 83(4), 764–770. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gean AD (1994). Imaging of head trauma. New York: Raven Press. [Google Scholar]
  16. Gordon NG (1972). The trail making test in neuropsychological diagnosis. Journal of Clinical Psychology, 28(2), 167–169. [DOI] [PubMed] [Google Scholar]
  17. Graham KS, & Hodges JR (1997). Differentiating the roles of the hippocampal complex and the neocortex in long-term memory storage: Evidence from the study of semantic dementia and Alzheimer’s disease. Neuropsychology, 11(1), 77–89. [DOI] [PubMed] [Google Scholar]
  18. Gupta R, Duff MC, Denburg NL, Cohen NJ, Bechara A, & Tranel D (2009). Declarative memory is critical for sustained advantageous complex decision-making. Neuropsychologia, 47(7), 1686–1693. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Heckers S (2001). Neuroimaging studies of the hippocampus in schizophrenia. Hippocampus, 11(5), 520–528. [DOI] [PubMed] [Google Scholar]
  20. Holdnack JA, Xiaobin Z, Larrabee GJ, Millis SR, & Salthouse TA (2011). Confirmatory factor analysis of the WAIS-IV/WMS-IV. Assessment, 18(2), 178–191. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Horecka KM, Dulas MR, Schwarb H, Lucas HD, Duff M, & Cohen NJ (2018). Reconstructing relational information. Hippocampus, 28(2), 164–177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kessels RP, de Haan EH, Kappelle LJ, & Postma A (2002). Selective impairments in spatial memory after ischaemic stroke. Journal of Clinical and Experimental Neuropsychology, 24(1), 115–129. [DOI] [PubMed] [Google Scholar]
  23. Konkel A, & Cohen NJ (2009). Relational memory and the hippocampus: Representations and methods. Frontiers in Neuroscience, 3(2), 166–174. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Leh SE, Schroeder C, Chen JK, Mallar Chakravarty M, Park MT, Cheung B, … Petrides M (2017). Microstructural integrity of hippocampal subregions is impaired after mild traumatic brain injury. Journal of Neurotrauma, 34(7), 1402–1411. [DOI] [PubMed] [Google Scholar]
  25. Maguire EA, Nannery R, & Spiers HJ (2006). Navigation around London by a taxi driver with bilateral hippocampal lesions. Brain, 129(Pt 11), 2894–2907. [DOI] [PubMed] [Google Scholar]
  26. Malec JF, Brown AW, Leibson CL, Flaada JT, Mandrekar JN, Diehl NN, & Perkins PK (2007). The mayo classification system for traumatic brain injury severity. Journal of Neurotrauma, 24(9), 1417–1424. [DOI] [PubMed] [Google Scholar]
  27. Monti JM, Cooke GE, Watson PD, Voss MW, Kramer AF, & Cohen NJ (2015). Relating hippocampus to relational memory processing across domains and delays. Journal of Cognitive Neuroscience, 27(2), 234–245. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Murray LL, Ramage AE, & Hopper A (2001). Memory impairments in adults with neurogenic communication disorders. Seminars in Speech and Language, 22(2), 127–136. [DOI] [PubMed] [Google Scholar]
  29. Palacios EM, Sala-Llonch R, Junque C, Fernandez-Espejo D, Roig T, Tormos JM, … Vendrell P (2013). Long-term declarative memory deficits in diffuse TBI: Correlations with cortical thickness, white matter integrity and hippocampal volume. Cortex, 49(3), 646–657. [DOI] [PubMed] [Google Scholar]
  30. Postma A, Kessels RP, & van Asselen M (2008). How the brain remembers and forgets where things are: The neurocognition of object-location memory. Neuroscience and Biobehavioral Reviews, 32(8), 1339–1345. [DOI] [PubMed] [Google Scholar]
  31. Postma A, Morel SG, Slot ME, Oudman E, & Kessels RPC (2018). Forgetting the new locations of one’s keys: Spatial-memory interference in Korsakoff’s amnesia. Experimental Brain Research, 236 (7), 1861–1868. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Rigon A, Turkstra L, Mutlu B, & Duff M (2016). The female advantage: Sex as a possible protective factor against emotion recognition impairment following traumatic brain injury. Cognitive, Affective & Behavioral Neuroscience, 16(5), 866–875. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Rigon A, Voss MW, Turkstra LS, Mutlu B, & Duff MC (2016). Frontal and temporal structural connectivity is associated with social communication impairment following traumatic brain injury. Journal of the International Neuropsychological Society, 22(7), 705–716. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Rigon A, Voss MW, Turkstra LS, Mutlu B, & Duff MC (2017). Relationship between individual differences in functional connectivity and facial-emotion recognition abilities in adults with traumatic brain injury. Neuroimage Clinical, 13, 370–377. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Royo NC, Conte V, Saatman KE, Shimizu S, Belfield CM, Soltesz KM, … McIntosh TK (2006). Hippocampal vulnerability following traumatic brain injury: A potential role for neurotrophin-4/5 in pyramidal cell neuroprotection. The European Journal of Neuroscience, 23(5), 1089–1102. [DOI] [PubMed] [Google Scholar]
  36. Rubin RD, Schwarb H, Lucas HD, Dulas MR, & Cohen NJ (2017). Dynamic hippocampal and prefrontal contributions to memory processes and representations blur the boundaries of traditional cognitive domains. Brain Sciences, 7(7), 82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Rubin RD, Watson PD, Duff MC, & Cohen NJ (2014). The role of the hippocampus in flexible cognition and social behavior. Frontiers in Human Neuroscience, 8, 742. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Schuff N, Woerner N, Boreta L, Kornfield T, Shaw LM, Trojanowski JQ, … Alzheimer’s Disease Neuroimaging I (2009). MRI of hippocampal volume loss in early Alzheimer’s disease in relation to ApoE genotype and biomarkers. Brain, 132(Pt 4), 1067–1077. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Schwarb H, Johnson CL, Daugherty AM, Hillman CH, Kramer AF, Cohen NJ, & Barbey AK (2017). Aerobic fitness, hippocampal viscoelasticity, and relational memory performance. Neuroimage, 153, 179–188. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Schwarb H, Johnson CL, McGarry MDJ, & Cohen NJ (2016). Medial temporal lobe viscoelasticity and relational memory performance. Neuroimage, 132, 534–541. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Serra-Grabulosa JM, Junque C, Verger K, Salgado-Pineda P, Maneru C, & Mercader JM (2005). Cerebral correlates of declarative memory dysfunctions in early traumatic brain injury. Journal of Neurology, Neurosurgery, and Psychiatry, 76(1), 129–131. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Sharp DJ, Scott G, & Leech R (2014). Network dysfunction after traumatic brain injury. Nature Reviews. Neurology, 10(3), 156–166. [DOI] [PubMed] [Google Scholar]
  43. Skidmore ER (2015). Training to optimize learning after traumatic brain injury. Current Physical Medicine and Rehabilitation Reports, 3(2), 99–105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Smith DH, Chen XH, Xu BN, McIntosh TK, Gennarelli TA, & Meaney DF (1997). Characterization of diffuse axonal pathology and selective hippocampal damage following inertial brain trauma in the pig. Journal of Neuropathology and Experimental Neurology, 56(7), 822–834. [PubMed] [Google Scholar]
  45. Spreng RN (2013). Examining the role of memory in social cognition. Frontiers in Psychology, 4, 437. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Squire LR (1992). Memory and the hippocampus: A synthesis from findings with rats, monkeys, and humans. Psychological Review, 99(2), 195–231. [DOI] [PubMed] [Google Scholar]
  47. Tate DF, & Bigler ED (2000). Fornix and hippocampal atrophy in traumatic brain injury. Learning & Memory, 7 (6), 442–446. [DOI] [PubMed] [Google Scholar]
  48. Tombaugh TN (2004). Trail making test A and B: Normative data stratified by age and education. Archives of Clinical Neuropsychology, 19(2), 203–214. [DOI] [PubMed] [Google Scholar]
  49. Tulving E, & Schacter DL (1990). Priming and human memory systems. Science, 247(4940), 301–306. [DOI] [PubMed] [Google Scholar]
  50. Vakil E (2005). The effect of moderate to severe traumatic brain injury (TBI) on different aspects of memory: A selective review. Journal of Clinical and Experimental Neuropsychology, 27(8), 977–1021. [DOI] [PubMed] [Google Scholar]
  51. Vespa PM, McArthur DL, Xu Y, Eliseo M, Etchepare M, Dinov I, … Hovda D (2010). Nonconvulsive seizures after traumatic brain injury are associated with hippocampal atrophy. Neurology, 75(9), 792–798. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Voss JL, Gonsalves BD, Federmeier KD, Tranel D, & Cohen NJ (2011). Hippocampal brain-network coordination during volitional exploratory behavior enhances learning. Nature Neuroscience, 14(1), 115–120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Voss JL, Warren DE, Gonsalves BD, Federmeier KD, Tranel D, & Cohen NJ (2011). Spontaneous revisitation during visual exploration as a link among strategic behavior, learning, and the hippocampus. Proceedings of the National Academy of Sciences of the United States of America, 108(31), E402–409. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Wang X, Gao X, Michalski S, Zhao S, & Chen J (2016). Traumatic brain injury severity affects neurogenesis in adult mouse hippocampus. Journal of Neurotrauma, 33 (8), 721–733. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Watson PD, Voss JL, Warren DE, Tranel D, & Cohen NJ (2013). Spatial reconstruction by patients with hippocampal damage is dominated by relational memory errors. Hippocampus, 23(7), 570–580. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Yee LT, Warren DE, Voss JL, Duff MC, Tranel D, & Cohen NJ (2014). The hippocampus uses information just encountered to guide efficient ongoing behavior. Hippocampus, 24(2), 154–164. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES