A standardized and reproducible behavioral framework to elicit multiple aspects of visual spatial attention in mice

Kayla Peelman; Nicole Allen; Jamin Ahn; Joseph Del Rosario; Kaitlin C Jacobson; Esther Kim; Anthony D Lien; Mercedes Lopez-Esteva; Lyndah Lovell; Kendell Worden; Yanlin Yang; James Zhuang; Bilal Haider

doi:10.1016/j.crmeth.2026.101514

. 2026 Jul 7;6(9):101514. doi: 10.1016/j.crmeth.2026.101514

A standardized and reproducible behavioral framework to elicit multiple aspects of visual spatial attention in mice

Kayla Peelman ¹, Nicole Allen ¹, Jamin Ahn ¹, Joseph Del Rosario ¹, Kaitlin C Jacobson ¹, Esther Kim ¹, Anthony D Lien ¹, Mercedes Lopez-Esteva ¹, Lyndah Lovell ¹, Kendell Worden ¹, Yanlin Yang ¹, James Zhuang ¹, Bilal Haider ^1,^2,^∗

PMCID: PMC13615504 PMID: 42413503

Summary

Understanding how neural activity gives rise to cognitive processes such as selective attention is a fundamental goal of neuroscience. An important but often overlooked advance toward this goal requires the development and sharing of rigorous and reproducible behavioral tasks across labs; this is particularly important given the recent surge in studies of cognition and perception in mice. Here, we developed a standardized training protocol for head-fixed mice to become experts in a psychometric visual contrast detection task in just 18 days of training. Experts detected stimuli at two distinct spatial locations for several hundred trials per day. As consecutive trials elapsed at either location, perceptual speed, accuracy, and contrast sensitivity improved—all hallmarks of spatial attention improving performance, as seen in primates. These findings validate the efficacy of this protocol to reveal multiple behavioral metrics indicative of selective spatial attention in mice, establishing a rigorous, reproducible tool for the community.

Keywords: selective attention, spatial attention, perception, visual system, mouse;psychophysics, contrast response function, reaction time, head-fixed behavior, protocol

Graphical abstract

Highlights

•
Detailed steps to train head-fixed mice in a psychometric visual spatial task
•
Operationalized protocol with task phases improves training efficiency and outcomes
•
In ∼18 days of training, naive mice become experts performing ∼400 trials/day
•
Spatial context improves performance, similar to effects of attention in primates

Motivation

Detailed reporting of training protocols for behavioral tasks is important for reproducibility and interpretability of the underlying neural mechanisms. This is because tasks can be learned and executed in different ways (even by experts), and subjects can be idiosyncratic in behavioral performance. It may thus be challenging to draw firm conclusions about the generality of neural mechanisms underlying behavioral performance across subjects, tasks, and species without standardized and reproducible training methods and rigorous performance tracking from naive to expert stages. Here, our motivation is to standardize and optimize training procedures for mice performing a selective visual spatial attention task.

Determining the relationship between neural activity and behavior is a major goal of neuroscience, and this requires standardized methods for training subjects in complex behavioral tasks. Peelman et al. detail a standardized behavioral protocol for training head-fixed mice in a visual task that elicits multiple signatures of spatial attention.

Introduction

One of the central goals of neuroscience is to determine the relationship between neural activity and behavior. This typically involves monitoring neural activity in animals while they perform a behavior of choice, then using analytical methods to relate the neural activity to behavioral features of interest. Such studies often report highly complex methods for neural activity measurements, data processing, data analysis, and statistical quantification, but with much less detail provided for the methodological decisions and procedures involved in the often equally complex behavioral tasks. When these details are provided, they often only describe end-stage, expert-level behavioral performance—not how animals were trained to achieve this state. This shortcoming in providing detailed behavioral training protocols alongside other reproducible deliverables of research (such as data and code reproducing the results) is a critical hindrance for the widespread replication of behavioral neuroscience studies. Indeed, recent large-scale, collaborative studies of visual perception have highlighted how standardizing and sharing behavioral protocols across labs improves reproducibility and interpretability of neural processes underlying behavior, a major step toward increasingly rigorous science.¹

Many fundamental studies of neural activity underlying behavior have focused on visual perception and how this process is affected by selective attention. Selective attention involves focusing processing to one aspect of the environment, while reducing processing of other aspects.²^,³ For example, selective attention to a particular region of visual space is accompanied by faster and more sensitive behavioral responses to stimuli in that region.²^,³^,⁴^,⁵ This selective improvement comes at the cost of slower responses and lower sensitivity to stimuli appearing elsewhere.⁶ For many decades, the neural basis of this process has been studied in great detail in head-fixed monkeys performing selective attention tasks.³^,⁵^,⁷^,⁸^,⁹^,¹⁰ Only very recently was it shown that head-fixed mice can also use selective visual attention to improve their visual processing.¹¹ Although some visual specializations of primates are absent in mice, they nonetheless share many fundamental organizational principles of visual processing, including retinotopy and hierarchical, multi-area feedforward, and feedback connectivity.¹²^,¹³^,¹⁴^,¹⁵^,¹⁶ Several groups have now developed tasks where expert mice show many signatures of visual spatial selection and attention.¹¹^,¹⁷^,¹⁸^,¹⁹^,²⁰^,²¹^,²²^,²³ These and other tasks for mice hold great potential for the use of advanced technologies²⁴^,²⁵ to study the neural basis of selective attention at scales that are more challenging in primates.

Detailed reporting of behavioral training protocols for sophisticated selective attention tasks in mice is particularly important for interpretability of the underlying neural mechanisms. Mice can be idiosyncratic during learning and behavioral performance,²⁶^,²⁷^,²⁸^,²⁹ and sophisticated tasks can be learned and executed in different ways by experts, with distinct consequences for neural activity underlying performance.³⁰ It is challenging to draw firm conclusions about the generality of mechanisms underlying selective attention across subjects, tasks, and species, without standardized and reproducible training methods and rigorous performance tracking from naive to expert stages. Thus, standardizing how mice are trained in selective attention tasks is critical for generating reproducible, interpretable inferences about this fundamental cognitive process.

Here, we detail a standardized behavioral protocol for training head-fixed mice to become experts in a visual task that elicits several signatures of spatial attention. With an average of 18 days of training, mice could report detection of small, localized visual stimuli that appeared at varying contrasts in one of two locations in the visual field. The stimuli appeared in blocks of consecutive trials at one location or the other. This alternating block structure elicited improvements in reaction time and accuracy as trials progressed at a given location, replicating our prior work²⁰; additionally, our current protocol elicited clear measures of attentional improvements in both monocular and binocular visual space. Furthermore, mice trained with this protocol also improved their contrast sensitivity across successive trials at a given spatial location, just like attentional improvements seen in monkeys and humans. These results further validate the efficacy of this protocol, establishing an efficient, high-throughput, and reproducible tool for the study of multiple aspects of selective spatial attention in mice.

Results

Behavioral task design and training phases

We previously developed a behavioral task for head-fixed, water-restricted mice that elicited both behavioral and neural signatures of visual spatial attention.²⁰ In that study, mice reached standardized expert-level performance, but without using a rigorously operationalized training protocol with standardized performance benchmarks during training. Here, we describe how standardizing and operationalizing key aspects of the training protocol into phases enables significantly faster learning to expert stages and consequently more robust measurements of several signatures of visual spatial attention that are repeatable across days and across subjects. The optimized protocol replicates our prior work and now improves training efficiency to increase throughput for the study of visual attention in mice.

The end goal of the task is to measure how expert mice use spatial attention to improve the speed and accuracy of visual perception; we will first briefly describe the expert stage of the task, to contextualize the subsequent descriptions of training stages. Expert mice report their perception of small, static visual gratings (shown at a range of contrasts; see STAR Methods) by licking for water rewards. Rewards are only available during a brief response window (1 s) while the stimuli are on the screen (Figure 1A); the first lick during the response window triggers reward delivery. Some trials are 0% contrast (blank) to measure the false alarm rate. Each detection trial is preceded by a mandatory quiescence period of randomized duration (0.5–15 s of blank screen; Figure 1A), where stray licks restart the trial; this incentivizes mice to withhold licking until stimulus appearance, then to lick rapidly upon stimulus detection. The key feature of expert-level task performance is detection of stimuli that appear in different spatial locations across blocks of consecutive trials. For example, in one block of consecutive trials, the stimuli appear in front of the mouse (Figure 1B, binocular block, blue). Then, without warning, the stimuli appear in a peripheral spatial location for the next block of consecutive trials (Figure 1B, monocular block, ochre). Aligning behavioral performance to the first trial of the block (when the stimulus switches from the expected to unexpected location) allows us to examine how perceptual speed, accuracy, and sensitivity change across consecutive trials in the block. If the block structure elicits spatial attention, we should observe that perceptual speed and accuracy are hindered on the first trial in the block (when attention is elsewhere), but then improve across consecutive trials at the new location.

Visual spatial detection task that elicits behavioral correlates of visual spatial attention

(A) Head-fixed, water-restricted mice reported stimulus detection by licking for water. All trials started with a quiescence period of randomized duration (0.5–15 s). Licking during this period restarted the trial. After stimulus onset, the first lick during a 1-s response window dispensed water (hit trial); failure to respond resulted in no reward (miss trial). On 20% of trials (bottom, blank trials), a 0% contrast stimulus was presented. Licking during these trials was recorded as a false alarm, while withholding licking was considered a correct rejection.

(B) At the expert stage, stimuli of varying contrast appeared in blocks for 10–30 consecutive trials at a single spatial location (binocular or monocular visual field). At the end of a block of trials in one location, the stimulus switched to the opposite spatial location without a cue.

Overall, averaged across all subjects trained with this protocol (N = 36), we observed clear behavioral effects of spatial attention (Figure S1A). Both detection speed and perceptual sensitivity improve significantly across consecutive trials of stimulus detection in peripheral (monocular) space, consistent with our previous study.²⁰ However, our prior study did not probe the effects of attention in the binocular visual field; here, we show the training protocol also elicits clear effects of visual attention in the binocular visual field (Figure S1B), as will be detailed later.

Training protocol with phases enables faster task expertise

The training protocol consisted of five distinct phases (Figure 2A), each designed to systematically develop task proficiency and ultimately reveal behavioral signatures of visual spatial attention. We briefly describe the phases here, then provide greater detail in subsequent sections. We refer the reader to the STAR Methods section “Definitions of trials, blocks, sessions, and days” for full description and definition of behavioral epochs.

Mice trained with a phased protocol become experts faster

(A) Schematic of sequential training phases: phase 0 (conditioning); phase 1 (shifting stimulus eccentricity); phase 2 (contrast shaping); phase 3 (switching stimulus locations); and optional phase 4 (introducing task-irrelevant stimuli). Attentional effects appear in expert mice from phase 3.

(B) Timeline of training phases, with average days per phase across mice. After surgical implantation, recovery, handling, and habituation plus water restriction (7 days total; last 1–2 days introduce water restriction), protocol trained mice took another 18 ± 7 days (mean ± SD) to become experts in the task (N = 36 mice). Phase 0 (3 ± 2 days, mean ± SD); phase 1 (7 ± 4 days); phase 2 (4 ± 3 days); phase 3 (4 ± 3 days); optional phase 4 (9 ± 11 days).

(C) Number of days each mouse spent in phases 0–3 (color), rank ordered by total training duration. 18 ± 7 days (mean ± SD).

(D) Mice trained in protocol (dark blue, N = 36) become experts in 18 ± 7 days, almost twice as quickly as mice trained without a standardized, phased protocol (gray, N = 11; 33 ± 20 days; p = 0.017, Kolmogorov-Smirnov test).

In phase 0 (operant conditioning), water-restricted mice learned that high-contrast stimulus appearance was associated with automatic delivery of water rewards (0.7 s after stimulus onset); this positively reinforces licking following visual stimulus onset. The stimulus is only shown in one location in phase 0 (at the vertical meridian in the binocular zone, defined as 0°) until mice show reaction times (first lick latencies) that are tightly aligned after visual stimulus onset and before the automatic reward delivery (anticipatory licking); thereafter, the automatic reward is discontinued, and the first lick during the visual stimulus (operant response) triggers reward delivery.

Phase 1 (eccentricity shifting) established stimulus detection at progressively greater distances from the center of the visual field. Eccentricity refers to the angular distance of a visual stimulus relative to the center of gaze, here in horizontal space. In mice, stimuli with eccentricity >40° away from the vertical meridian (defined as 0°) appear in the monocular visual field. We gradually moved the stimulus location in 10°–20° increments toward the monocular field (up to 70°–90° eccentricity), with progression determined by performance criteria.

In phase 2 (contrast shaping), stimulus position remained fixed at 70°–90°, and they were shown at varying contrasts. This ensures mice learn to detect stimuli of different intensities and allows assessment of psychometric performance spanning the detection threshold.

Phase 3 (location switching) introduces contrast-varying stimuli in two different locations (either 0° or the peripheral location), in blocks of 10–30 consecutive trials at a given spatial location. During each behavioral session, trial blocks switch from one location to the other. Phase 3 allows concurrent assessment of both psychometric performance and spatial attention effects elicited by block switching.

Finally, in mice eventually used for neural recordings, we introduced an optional phase 4 (task-irrelevant stimuli). Mice continued performing the task exactly as described in phase 3, but in the presence of low contrast, task-irrelevant vertical bars flashed across the visual field. These task-irrelevant stimuli are used to assess attention-related changes to neural spatial receptive fields.²⁰ This stage is not required to assess the behavioral effects of spatial attention induced by location switching.

We hypothesized that this structured, phased approach would expedite task learning compared to our prior training methods. Indeed, 92% of mice designated for training with the phased protocol (36/39; see STAR Methods for inclusion criteria) became experts in just 18 ± 7 days (Figures 2B and 2C; through phase 3). Compared to prior cohorts of non-protocol trained mice, this was a 46% reduction in training time (Figure 2D; p = 0.017, Kolmogorov-Smirnov test) and with significantly less variance in total training time (N = 11 mice, p = 0.033, Brown-Forsythe test). Protocol-trained experts performed 21 ± 6 spatial location switches per day and performed 369 ± 78 trials (mean ± SD), providing hundreds of thousands of standardized behavioral trials in a large cohort for detailed analysis of attentional effects, as will be shown later. In the next sections, we detail the quantitative criteria used for progression through each of the training phases.

Operant conditioning establishes stimulus-reward association

Mice learned the association between visual stimuli and rewards through operant conditioning (phase 0; Figure S2). During this phase, after a randomized blank screen period (1–15 s, see STAR Methods), static horizontal Gabor gratings (65%–85% contrast) appear for 2 s in the center of the binocular visual field (at 0° azimuth). Water rewards were delivered automatically 0.7 s after stimulus onset to encourage mice to associate the visual stimulus with water delivery. Mice showed evidence of learning the stimulus-reward association with a progressive shift of first lick latencies from times after reward delivery (Figures S2A and S2B; 0.93 ± 0.05 s, first day of phase 0 training) to times before reward delivery (Figure S2B; 0.55 ± 0.08 s, N = 36 mice, median ± IQR/2, final day of phase 0 training, p = 2.56e−7). They concurrently showed a higher probability of trials with lick responses across the 3–9 days of phase 0 training (Figures S2C and S2D). This shift in the timing and probability of licking indicates a transition from reward-driven licking to visual-driven licking. On the last day of phase 0, automatic reward delivery was discontinued, and reward was only delivered if a lick occurred during visual stimulus presentation, thus directly reinforcing the visually elicited action (operant conditioning). Mice advanced to phase 1 when these lick times occurred consistently before 0.7 s with low trial-to-trial variability (Figures S2C–S2F). Thus, at the end of phase 0, licks report perceptual detection of the visual stimulus itself.

Stimulus detection across the visual field and at varying contrasts

In phase 1, we shortened the response window (to 1 s) to encourage fast reaction times and gradually moved the stimulus from the central to the peripheral visual field (Figure 3A). As shown in our previous study and several others,³¹^,³²^,³³ mice show greatest visual sensitivity in the binocular visual field (spanning ±40 degrees of central visual space) (Figures S3A and S3B). Accordingly, the stimulus becomes more difficult to detect as it shifts in eccentricity away from the vertical meridian. This necessitates gradual, stepwise introduction of the stimuli in monocular locations, with concurrent assessment of performance as a function of stimulus position. The stimulus is shown at a fixed position in a block of trials (typically 50–100) and then shifted by 10°–20°, pending performance metrics (hit rate >75%, d’ > 1). It is important to enforce low false alarm rates in phase 1 (captured by d’) before moving the stimulus position. In an example mouse (Figure 3B), stimuli placed in the binocular zone resulted in high detection accuracy (hit rate [HR] > 80%). However, as the stimulus shifted in eccentricity across sessions, accuracy often abruptly decreased; further, if stimulus eccentricity shifted too suddenly (i.e., before achieving high performance at a given location), mice often disengaged from the task or would not perform consistently at more eccentric stimulus positions. However, with repeated exposure at these locations, performance stabilized and improved. A similar trend was observed across the population (Figure 3C): performance remained consistently high using gradual, stepwise shifts in stimulus eccentricity. Importantly, these improvements were not accompanied by significant increases in the false alarm rate (Figure S3C). These results highlight the importance of gradually shifting the eccentricity of the stimulus for ensuring high detection performance at these more difficult spatial locations.

Shifting stimulus eccentricity from the central to peripheral visual field

(A) Grating stimulus is presented at a fixed location for a block of trials. Across blocks of trials, grating location is slowly moved in increments of 10°–20° toward the monocular field of view.

(B) Performance in an example mouse. Slow progression of stimulus from the binocular to monocular visual field keeps performance high. False alarm rate continues to decrease while hit rate remains stable.

(C) Population of mice trained in phases (N = 36). Data plotted are mean ± SEM across mice. On the final day of phase 1 where the stimulus is at 70°, the hit rate is 82% ± 12% (mean ± SD) and false alarm rate is 27% ± 11%.

Phase 2 training introduced stimuli of different intensities and measured psychometric detection sensitivity (Figure 4A). After establishing high detection performance for high-contrast stimuli in the far periphery (monocular space; ∼70°), we next systematically varied the range of contrast levels across sessions while keeping the stimulus fixed at this location (4–6 different contrasts per session; 100–150 trials per session). This has two purposes: (1) it prevents behavioral ceiling effects (i.e., an inability to measure attentional improvements beyond maximal performance at high contrast), and (2) it ensures that mice are performing around their detection thresholds, creating conditions that facilitate task engagement so that attention could improve the speed and probability of attaining rewards. During the early sessions of phase 2 (first 25% of sessions), a range of medium to high contrasts appeared (e.g., 18%, 33%, 65%, 85%), resulting in high performance (>80% correct) across all stimuli (Figure 4B, orange). In the subsequent sessions, when performance at the highest contrasts exceeds 80% or when responses at the lowest contrast rise significantly above the false alarm rate, the contrast range is shifted downward by one contrast step, ensuring that responses sample the full dynamic range. As a result, the lowest contrasts in early sessions become the highest contrasts in the later sessions. In late sessions of phase 2 (last 25% of sessions), this adaptive approach reveals psychometric performance that spans from the false alarm rate to maximal HRs (Figure 4B, blue circles). In addition, late phase 2 is characterized by improved performance at intermediate and low contrasts (Figures 4B and 4C). Population analysis revealed a significant reduction in contrast detection threshold in late versus early phase 2 training (Figure 4D). These effects were not due to estimating thresholds from psychometric fits or by using non-overlapping contrast ranges: in 25/34 mice, hit rates for an exactly shared “reference” contrast value (33%) significantly improved in late vs. early sessions (Figure 4E). Similarly, the slope of the psychometric curve became steeper from early to late phase 2 (Figure 4F). This steeper slope indicates improved sensitivity, such that smaller changes in stimulus contrast around threshold produce larger changes in performance. At the same time, false alarm rates did not significantly change across phase 2 training (Figure 4G). These improvements of perceptual accuracy validate the effectiveness of our structured, multi-phase training approach for optimizing psychometric visual detection. Once mice reached this stage of high-quality psychometric performance in peripheral stimulus space, we introduced stimulus location switching to assess spatial attentional effects.

Gradual shaping improves psychometric contrast detection at monocular locations

(A) Schematic of phase 2: contrast shaping. Grating stimulus stays in the monocular location (70°) but appears across a range of contrasts (4–6 different contrasts, not including 0%).

(B) Psychometric performance during early (orange) vs. late (blue) phase 2 training for one example mouse. Note that in early contrast shaping, performance is high for all contrasts, while in late contrast shaping, the mouse exhibits psychometric performance at a range of contrasts that evenly sample the psychometric range (N = 1 mouse; early R²: 1.00; late R²: 0.90). Dashed lines indicate contrast threshold for early (orange) and late (blue) curves. Curves fit using a Boltzmann equation (see STAR Methods) throughout figure.

(C) Early (orange) vs. late (blue) psychometric performance during phase 2 of training for entire population (N = 34 throughout figure unless otherwise noted; early average population R²: 0.91 ± 0.21; late R²: 0.91 ± 0.17). Note that two mice were omitted because of only 1 day of phase 2 training, preventing “early” and “late” comparisons.

(D) Contrast threshold decreases from early to late phase 2 training (early: 18.6% ± 9.6%; late: 5.8% ± 2.3%; median ± IQR/2; *p =* 4.9e−5, signed rank test throughout figure unless otherwise noted). Each gray line indicates an individual mouse’s change in contrast threshold from early to late phase 2. Error bars indicate median ± interquartile range (IQR) throughout figure.

(E) Hit rate improves at lower contrasts across training in phase 2. Threshold contrast was established from psychometric fits during early training (50.0% ± 6.8% accuracy), and hit rate at that same contrast was calculated during late phase 2 training (72.6 ± 13.3% accuracy at same contrast value; *p =* 1.1e−4). Hit rate significantly improves for stimuli of the exact same contrast (33%) in late (blue, 0.84 ± 0.06) versus early (orange, 0.77 ± 0.1) sessions (*p =* 0.013, signed rank test, median ± IQR for both, N = 25 mice).

(F) The local slope of the psychometric curves around threshold significantly increases from early (2.48 ± 0.64) to late (4.79 ± 1.26) sessions in phase 2 training (*p =* 1.15e−6).

(G) False alarm rate (FAR) does not significantly increase across phase 2 training (early: 23.3 ± 8% FAR; late 20.7 ± 8% FAR; *p =* 0.845).

Stimulus location switching elicits spatial attentional effects on behavior

Stimulus location switching (phase 3) provoked behavioral effects consistent with spatial attention, in both binocular and monocular visual fields. We ensured psychometric performance throughout the visual field during location switching. Unlike monocular detection, which requires gradual shaping for high psychometric performance, binocular detection is both intrinsically easier and more contrast sensitive (Figure S3), so we can immediately re-introduce binocular stimuli at different contrasts, but at a range one step lower than the monocular stimuli. This lower set of contrasts spans the full dynamic range from chance level detection to maximal performance (Figure 5A). Just as in phase 2, the mandatory intertrial interval without licking (drawn from an exponential distribution ranging from 1 to 8 s, on average), a 20% probability of blank trials, and the randomized selection of trial-by-trial stimulus contrast all made the exact timing and intensity of stimuli unpredictable (see STAR Methods, “task parameters”). The stimuli appeared in a given location for a consecutive block of trials, typically 10–20 (randomized) depending on performance history per subject (see STAR Methods “Training phase 3: Stimulus location switching”), with twice as many trials shown in the more difficult monocular location. During location switching, we measured high-quality psychometric performance in both locations for all mice trained with the phased protocol (Figure 5A; binocular psychometric curve fit quality R² = 0.97 ± 0.05; monocular R² = 0.95 ± 0.12, mean ± SD, N = 36 mice). This defines expert-level task performance.

Stimulus location switching reveals behavioral effects consistent with spatial attention in both binocular and monocular spatial locations

(A) Psychometric performance is observed in all mice (N = 36) for both binocular (left, blue) and monocular (right, ochre) spatial locations (Boltzmann fits; binocular R² = 0.99; monocular R² = 0.99; figure error bars represent mean ± SD across 759 days in N = 36 mice, performing 15,617 total block switches, 95,295 total binocular trials, and 180,617 total monocular trials; on average mice executed 21 ± 6 block switches and performed 369 ± 78 trials per day. Each mouse contributed an average of 21 ± 11 days of data; mean ± SD). All eligible days in phases 3 and 4 were analyzed (no sub-selection).

(B) Sensitivity improves significantly as a function of trial number in block for monocular stimuli (right, ochre; *p =* 6.0e−5, linear mixed-effects model throughout figure; see Table S1). Binocular sensitivity improves after trial 1 but not significantly (left, blue; *p =* 0.64). Data are plotted as mean ± SEM across days. Note different scales for ordinate.

(C) Reaction times become significantly faster as a function of trial number in block for both binocular (left, blue; *p =* 2.9e−30) and monocular stimuli (right, ochre; *p =* 2.9e−11). Data are plotted as mean ± SEM across days.

As mice detected stimuli spanning psychometric thresholds, switching the stimulus location across blocks elicited attentional improvements of overall detection sensitivity (d’) across successive trials (Figure 5B). Sensitivity improved after the first trial in either location, but significantly for monocular stimuli (p = 6e−5, linear mixed-effects model, n = 5,313 observations from phases 3 and 4 in 36 mice; see STAR Methods and Table S1 for model details). Simultaneously, reaction times significantly decreased across these same trials in both locations (Figure 5C, monocular: p = 2.9e−11; binocular: p = 2.9e−30). Restricting to threshold contrasts revealed similar trends on improved sensitivity and reaction times within block (Figures S1F–S1G). These effects did not emerge only with learning or extensive practice in phase 4. Restricting the analysis to all phase 3 days alone (20% of Figure 5 dataset) showed that the effects were already present from the start of location switching (Figures S1C–S1E), and restricting only to the first day of phase 3—the very first instance of block switching—already revealed significant improvements for reaction times in both locations (Figure S1E caption text). Additionally, there was no significant correlation between the number of days elapsed across phases 3 and 4 and detection accuracy on the first trial of each block; this suggests that the timing of the block switch remains unpredictable and unexpected across days, since the sensitivity of detecting that first stimulus in the new location is not improving with learning across days (Figure S1H). Across these same days, there was also no strengthening of the within-block d’ improvement (Figure S1I). There was also no significant relationship between the number of days a mouse spent in any particular training phase and the likelihood of showing attentional effects. Averaged across phases 3 and 4, increased sensitivity and faster reaction times across trials were evident in the majority of mice trained in the protocol: 83% exhibited at least one day with significant behavioral effects of attention during phases 3 and 4, and 61% of individual subjects showed significant effects in the average of all their phases 3 and 4 sessions (no selection criteria; Figure S4).

Importantly, these behavioral improvements were not associated with changes in eye position or pupil area across trials following block switches (Figure S5), nor with an increase in false alarms during location switching (phase 2 = 23% ± 10%; phase 3 = 23% ± 6%; phase 4 = 24% ± 6%; mean ± SD; p = 0.350, Kruskal-Wallis test). Together, these findings confirm that just like prior versions of this same task,²⁰ location switching across blocks reliably elicits spatial attentional effects on improved speed and accuracy of stimulus detection, even from the very first sessions and days of experience; the magnitude of our effects is consistent with prior reports of reaction time and accuracy benefits with spatial attention in mice and consistent with the magnitude of effects in studies of spatial attention in monkeys and humans (Table S3). This new phased protocol elicits attentional effects in both spatial locations, all while mice detect stimuli at a range of contrasts throughout the visual field; this new feature of the training protocol allowed us to rigorously assess the potential for attentional effects on psychometric contrast sensitivity, examined next.

Standardized training protocol reveals improved contrast sensitivity with attention

We wondered if this cohort of expert mice could reveal additional effects of attention that were not explicitly designed into the protocol. Inspired by observations in monkeys and humans,³⁴^,³⁵ we assessed if attentional improvements in detection speed and accuracy were accompanied by heightened sensitivity to contrast. Influential models of attention³⁶^,³⁷ show that the overall increase in responses to stimuli can result from distinct changes to the contrast response curve, each with different effects on contrast sensitivity (Figures S6A–S6G). If overall increased detection accuracy arises from the response probability to every stimulus (hit rate) being multiplied by a constant gain factor (response gain), sensitivity increases mainly at high contrasts, but with little change in the threshold contrast value (i.e., the contrast detected correctly 50% of the time, c₅₀; Figures S6E–S6G); if instead overall detection accuracy increases because the entire response curve shifts to the left (subtractive change, or contrast gain), sensitivity improves mostly at lower contrasts, because lower contrast stimuli are now detectable ≥50% of the time (c₅₀ decreases; Figures S6B–S6D). Either model can account for the same overall improvements of detection accuracy (Figures S6C and S6F), but it is not known if contrast sensitivity of mice could be modulated by block switching in two discrete locations of space, nor which model best explains data in our task. To test this, we aggregated and aligned all the trials following block switches in protocol trained mice, selecting days with overall increasing d’ across successive trials following block switches (monocular: N = 34 mice, 234 days, 808 block switches; binocular: N = 35 mice, 254 days, 834 block switches).

We generated psychometric curves as a function of trials elapsed after the block switch. Recall that contrasts on every trial were randomized—trial 1 could be 10% contrast in one block, 18% contrast in another, etc. Thus, aggregated across all blocks and mice, the first trials of all the block switches collected responses from all contrasts. The responses for all these 1^st trials following the block switch (spanning all contrasts) were sorted to construct a population level psychometric curve (Figures 6A–6D, black); the same procedure was done for the subsequent trials 2–5 (Figures 6A–6D). These block-switch aligned psychometric curves were constructed separately for each spatial location, showing proportion correct as a function of contrast and of trials elapsed across the block.

Contrast sensitivity improves after the stimulus appears in a new location

(A) Psychometric curves for binocular detection showing hit rate as a function of stimulus contrast for trials following a block switch. Data from days with overall increasing d’ across successive trials following block switches in phases 3 and 4 were aggregated for each trial number (1–5) after the switch into the binocular condition. Data are plotted as mean ± SEM across days throughout figure.

(B) Contrast threshold (c₅₀), derived from trial-specific psychometric curves, was measured for binocular blocks (see STAR Methods for defining contrast threshold). Following a switch to the binocular location, contrast threshold decreased significantly as a function of trial number in the block (linear mixed-effects model throughout figure, see Table S1, *p =* 1.17e−6.). See Figure S6 for contrast gain versus response gain model predictions on contrast sensitivity and correlation to real data.

(C) The slope of the contrast response function increased significantly as a function of trial number within block (*p =* 0.032).

(D) Psychometric curves as in (A), but when the stimulus switched to the monocular location.

(E) Following a switch to the monocular location, contrast threshold significantly decreased as a function of trial number in the block (*p =* 2.38e−6).

(F) The slope of the contrast response function increased but not significantly as a function of trial number within block (*p =* 0.16).

This analysis revealed that, indeed, spatial block switching also improved contrast sensitivity across successive trials within block. Contrast thresholds (c₅₀) and slopes of the psychometric function at threshold were analyzed as in a prior study.³⁴ Contrast thresholds significantly decreased across trials in both spatial locations (Figures 6B–6E; linear mixed-effects model: binocular: p = 1.2e−6; monocular: p = 2.3e−6). Similarly, the slope of each trial-derived psychometric curve increased as a function of trial number within a block, significantly for binocular stimuli (Figures 6C–6F; binocular: p = 0.032; monocular: p = 0.158); similar trends were observed using multiple methods of estimating slope around C₅₀. We then compared the entire matrix of [hit rates × contrast × trial number within block] to predictions from either contrast gain or response gain models. Our data were significantly correlated with the prediction from contrast gain, not response gain, in both monocular (p = 0.037; Figures S6H–S6N) and binocular locations (p = 0.011; Figures S6O–S6U). This did not have to be the case since either model could account for the same overall d’ improvements we used to sub-select days (Figures S6C–S6F). The selection of days with increasing d’ within block likely identified the highest performance sessions that were least variable across mice, since the trends were overall consistent but less apparent without any performance indexing (Figure S7). These results establish that contrast sensitivity in mice also improves in a manner consistent with effects of spatial attention in monkeys³⁸ and humans.³⁹ These findings appeared in a large cohort of mice trained in a standardized protocol that was optimized to reveal effects of attention on two other behavioral correlates—improved reaction times and accuracy. These results on improved contrast sensitivity further validate our protocol’s efficacy for instantiating multiple attentional effects on the speed, accuracy, and sensitivity of visual perception in mice.

Discussion

Here, we present behavioral data from a standardized protocol in which head-fixed mice become experts in a challenging psychometric visual task that elicits several signatures of spatial attention. Our protocol was explicitly designed to reveal changes in visual detection accuracy and reaction time as readouts of attentional improvements. We measured how these two quantities changed following the appearance of the task-relevant visual stimulus in an unexpected spatial location. Both quantities improved as consecutive detection trials elapsed at the newly relevant spatial location, consistent with well-known effects of spatial attention in monkeys and humans.²^,³^,⁶^,⁸^,³⁴^,⁴⁰^,⁴¹^,⁴² Our protocol showed significant improvements from our prior training procedures: task acquisition, expert-level performance, and attentional effects were repeatable across mice, and these were achieved in significantly fewer training days. Furthermore, by rigorously enforcing psychometric contrast detection in two spatial locations, expert mice trained with the protocol also showed improved visual contrast sensitivity across successive trials at a given spatial location, consistent with findings in monkeys and humans. Taken together, these findings establish and validate an efficient, high-throughput, and reproducible behavioral protocol for the study of spatial attention in mice, adding a valuable community resource for replicable behavioral neuroscience research.

The first advance of our protocol is that training times were fast and reliable. Mice needed just 18 days of training on average to achieve expert-level performance (defined as high-quality psychometric detection during location switching), cutting in half the required training days from our prior work. This time to expert-level performance is faster than other selective visual attention tasks in mice that report training times, ranging from 5 to 8 weeks²¹ up to 5–7 months.¹⁷^,⁴³ Further, the variability of training days to reach expert stage was markedly reduced versus our prior work that used a less operationalized training regime for the same task. Predictability of training durations is particularly important for experiments with time-sensitive variables (such as opsin expression, developmental milestones, studies of aging, chronic implantations, or studies with mouse models of neurological disorders that show age-related symptoms).

The second advance of our protocol is high efficiency of task acquisition and expert-level performance. Only 3% of mice (1/39) failed to reach expert level due to poor behavioral performance. Expert mice performed nearly 400 trials per day, and switched the location of visual detection 21 times per day. Several recent studies of visual attention in mice provide some comparison points for our protocol efficacy. In one study of audiovisual spatial attention, 37% (of 51 subjects) reached expert-level performance.¹⁷ In another study, around 18% of mice were removed from training for failure to learn the task within 3 months,¹¹ but the remaining mice show high incidence of attentional effects within subjects. Another study of visual attention in freely moving mice showed between 11% and 32% exclusion of subjects due to poor task performance.¹⁹ Our protocol efficacy compares favorably with these prior reports. However, it is important to note that we purposefully defined expert-level performance solely from psychometric criteria during location switching—not by the presence (or absence) of attention effects. This was to avoid experimenter bias in shaping the very effects we set out to test. Expert-level performance could certainly be achieved and maintained without requiring attention, since our task did not overtly cue or direct it. Nevertheless, location switching elicited multiple behavioral metrics consistent with spatial-attention-mediated improvements in performance.

Our gradually phased protocol with progressive increases of difficulty and session-by-session monitoring helped mitigate biases wherein mice favor trials in the binocular visual field, since this region is inherently much more perceptually sensitive.²¹^,²²^,³¹^,³²^,³³ Recent studies of visual selection and attention show that mice can adopt idiosyncratic biases for visual field preferences,¹⁸^,²¹ necessitating subject exclusion/grouping to average across mice with similar biases. Our protocol revealed clear population-level evidence for behavioral improvements in detection speed and contrast sensitivity in both the monocular and binocular visual fields, while detection accuracy improved in both locations but significantly in the monocular visual field. Importantly, these population-level findings on attentional improvements of accuracy, reaction times, and contrast thresholds were evident from the same interleaved behavioral sessions (with sequential location switching across both locations), without sub-selecting only those subjects with statistically significant behavioral effects or only selecting trials from one location or the other. Prior studies of selective spatial attention in mice that briefly cued the relevant visual hemifield¹¹ established that this improves psychometric sensitivity to ensuing stimulus orientation changes. To our knowledge, our findings are the first to show in mice that spatial attention in either the central or the peripheral visual field is associated with improved contrast sensitivity along with faster and more accurate detection. This did not necessarily have to be the case, since overall improvements to d’ can occur with no change to contrast sensitivity (Figure S6); these findings are most consistent with the contrast gain model of attention; in humans, contrast gain results when the attentional field is larger than the stimulus field.³⁹ Exploring these topics now seems approachable in mice.

Our training protocol offers several advantages for ease of implementation and scalability. First, mice were motivated only through water restriction, reducing complications associated with food restriction. Water delivery is easy to implement, calibrate, and measure in real time and provides an easy way to equalize motivation across subjects (body weight criteria provide an excellent proxy for hydration level and thirst).¹^,²⁶ Second, we did not use overt punishments to shape behavior, e.g., air puffs and white noise.¹¹^,¹⁷^,²¹ This simplifies the construction of the task and apparatus and does not introduce interpretational complications of both appetitive and aversive learning in task acquisition and performance. False alarms stayed at low levels in experts, with no need for aversive shaping. Third, by starting operant conditioning for visual stimuli in the most sensitive visual region (binocular space), then gradually introducing increasingly difficult aspects of the task (stimuli in peripheral space, at varying contrast levels, then switching locations), this ensures performance metrics at each training stage can be interpreted according to controlled introduction of new task variables, apart from concurrent learning of the basic stimulus-reward relationship. Fourth, mice learn and perform the task in a semi-enclosed plastic tube, where no locomotion is possible. This is important because locomotion introduces widespread neural modulation across the brain, increases eye movements, impacts perceptual performance,⁴⁴^,⁴⁵ and can dramatically alter the strength of visual responses, all potentially complicating the interpretation of attention effects.¹⁸^,⁴⁶^,⁴⁷^,⁴⁸ In this way, our task design aligns more closely with head-fixed, stationary conditions in monkeys,¹⁰^,⁴⁹ facilitating comparisons of attentional modulation across species.

There are several design choices and interpretational considerations for our task and its relationship to other tasks of visual perception and attention in mice and primates. First, our task is a detection task, not a discrimination task. Detection tasks where subjects report their perception of stimulus presence have a long history of utility for studying visual perception in humans⁵⁰^,⁵¹ and for studying the neural basis of perceptual processes across species⁵²^,⁵³^,⁵⁴^,⁵⁵^,⁵⁶^,⁵⁷^,⁵⁸; however, discrimination tasks can probe subjective perceptual judgments comparing stimulus feature attributes,⁵⁹^,⁶⁰^,⁶¹ which our task cannot assess. Second, attentional processes are clearly tied to both reward expectancy and spatial expectancy, processes that are difficult to fully disentangle and explain with a purely sensory manifestation of attentional processes.⁶²^,⁶³ Indeed, a recent study in mice with uncued, spatially blocked trials shows that behavioral improvements across trials in many ways resemble an updating of a spatial prior (or expectation) in a Bayesian framework.⁶⁴ Independently manipulating probabilities of stimuli and rewards⁶⁵^,⁶⁶^,⁶⁷ may be able to shed light on these factors in future studies. Third, our task does not cue attention trial-by-trial, a common feature of tasks in primates. Trial-by-trial cueing is an important and rigorous manipulation, but it is not the only way to study visual attention. Changing spatial probabilities across blocks—without trial-by-trial cueing—is also a potent driver of spatial attentional allocation in humans⁶⁸^,⁶⁹^,⁷⁰ and in other attentional tasks in mice.¹⁹^,²¹ Many fundamental studies of attention in monkeys also use block designs like ours with a high probability of stimulus events at the relevant location and without explicit trial-by-trial cueing during the block.³⁸^,⁶⁰^,⁷¹^,⁷²^,⁷³ Although we do not explicitly cue the location of attention, there are nonetheless some points of comparison to our task. In our task, the trials within a block can all be thought of as “validly” cued by the stimulus on the previous trial in the same location—except for the first stimulus in the block. For this first trial, the immediately preceding stimulus appeared in the other location, and thus the first trial of the block switch is “invalidly” cued for the new stimulus location. This predicts a performance decrement on the first trial of the block, just like an “invalid” cue; this is precisely what we observed. The performance decrements on this first trial then dissipate with subsequent “valid” stimuli appearing at the same location within block. This is consistent with an interpretation of a gradual, trial-by-trial attentional shift to the new location; this was evident even from the first day of block switching and did not suddenly emerge or improve with practice days. Lastly, it bears mentioning that in a prior study using this very same task, we observed several neural correlates in V1 (increased gain and reduced noise) during perceptual improvements that were spatially specific, extended to task-irrelevant stimuli (the bars in phase 4), and could not be explained simply by reward expectancy or motor preparation.²⁰ These findings echoed those using task-irrelevant stimuli in monkeys during spatial attention tasks.⁷⁴^,⁷⁵

There are several behavioral factors that could be considered in the future to improve training efficiency. For one, we did not account for variations in learning speed across animals, although these were evident (Figure 2C). These variations may arise from many factors including strain differences,⁷⁶^,⁷⁷ breeding paradigms,⁷⁸ animal sex,²⁸ and environmental factors.⁷⁹ All of these could be further analyzed and controlled. Second, mice were trained on a reverse-light cycle and at approximately the same hour of day within subject, but we did not examine if circadian effects explain subject-to-subject learning variability and performance,⁸⁰^,⁸¹ a topic for future consideration. Third, trainer sex⁸² and trainer experience can also impact training variability, although both novice and experienced trainers of either sex trained mice of either sex to expert-level performance using the protocol.

Attention is a fundamental cognitive process that influences our vision and perception; our mechanistic understanding of this has come from decades of foundational studies in primates. Our protocol provides an efficient and reproducible way to study spatial attention behavior in head-fixed mice, permitting future use of cell-type-specific perturbations, advanced optical and electrical recordings, and use of transgenic mouse models of neurological disorders, many of which show deficits and impairments of attention. By pairing these techniques with rigorous and reproducible procedures for training behavioral tasks, and with careful consideration of behavioral and species-specific differences, mice can provide detailed insights into the neural basis of attention and more clearly delineate conserved versus ethologically distinct neural mechanisms of attention in mammalian sensory systems.

Limitations of the study

Our training protocol carries several inherent limitations resulting from task design choices. A first limitation is that visual stimuli do not appear in symmetric positions in each visual hemifield across blocks of trials,¹¹ as typically positioned in primate studies.³⁴ Our choice was for several reasons: (1) to exploit the greater visual sensitivity of binocular visual space during the initial operant conditioning phase; (2) so that we could study the differences in perception (and its improvement by attention) in these two functionally distinct regions of visual space; and (3) so that during recordings, stimulus-evoked neural activity in monocular blocks is initially localized to one cerebral hemisphere. Much evidence shows that the binocular visual field in mice is specialized and behaviorally privileged, even in the absence of a fovea,²¹^,²²^,³¹^,³²^,³³ as evident in our data measuring psychometric contrast sensitivity in both binocular and monocular space in the same sessions. Our task provides a standardized and rigorous way to examine the neural basis of these visual field differences. A second limitation is that our task can only examine the progression of attentional effects across successive trials following the first appearance of the visual stimulus in the unexpected location (block switch). Since the duration of each intertrial interval is randomized and unpredictable (see STAR Methods), we do not have explicit events (such as an orienting cue) that align the onset of attention in the exact same way on every single trial. Our task was explicitly designed to avoid the potential complications of rapid visual cueing (masking⁸³ or stimulus rebound⁸⁴^,⁸⁵) and rapid successive visual stimuli (adaptation⁸⁶; this reduces detection sensitivity) and to instead probe attentional allocation based on statistical changes in the visual environment. Multiple behavioral readouts in our task indicate that mice took several trials to shift their attention, likely accumulating evidence of stimulus location and successive reward outcomes across trials within block. This slow accumulation across trials (with randomized intervals spanning seconds to tens of seconds) clearly revealed improvements in perceptual speed and accuracy but constrains temporal examination of the onset and duration of attentional effects. Modifications to our task with visual cueing of the relevant spatial location could be implemented to study temporally aligned attentional effects.¹¹ Importantly, the reaction time improvements due to spatial block switching were evident from the very first day of exposure to phase 3 and did not emerge only with extensive practice in phase 4, nor did the effects lessen across training phases. However, we cannot exclude that a learning process may contribute to stabilization of the improvements in accuracy associated with block switching. A third limitation, common to all detection tasks, is the difficulty in interpreting failed detection trials, relative to two choice tasks.¹⁹^,⁸⁷^,⁸⁸^,⁸⁹ However, our use of blank stimulus trials with the exact same temporal structure as true stimulus trials shows that false alarms did not increase across successive trials in the block, while hit rates, reaction times, and contrast thresholds all improved—consistent with attentional improvements associated with greater visual sensitivity.⁴¹ Furthermore, our prior studies using this same contrast detection task that spans the psychometric range show that trial-by-trial activity in primary visual cortex (V1) is necessary for performance,²⁰^,³³^,⁹⁰ similar to findings in other visual contrast detection tasks in mice.⁵⁵^,⁹¹ The specific brain areas and cell types critical for learning and executing⁹²^,⁹³ this task are an important topic for future studies.⁹⁴

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Bilal Haider (bilal.haider@bme.gatech.edu).

Materials availability

This study did not generate new, unique reagents.

Data and code availability

•
All data structures used to generate each figure are publicly available as of the date of publication at DOI: 10.6084/m9.figshare.30375733 and linked from the corresponding author’s institutional webpage.
•
All original code used to generate each figure are publicly available as of the date of publication at DOI: 10.6084/m9.figshare.30375733 and linked from the corresponding author’s institutional webpage.
•
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We thank members of the Haider lab for feedback. This work was supported by the Alfred P. Sloan Foundation’s Minority Ph.D. (MPHD) Program Fellowship (to J.D.R.), the National Eye Institute (F31EY033691 to K.P.) and the National Institute of Neurological Disorders and Stroke (T32NS096050 to K.P. and R01NS109978 and RF1NS132288 to B.H.), the Simons Foundation (grant no. SFARI 600343 to B.H.), and the National Science Foundation Graduate Research Fellowship Program (DGE-2141064 to K.C.J.).

Author contributions

Conceptualization, K.P. and B.H.; data curation, K.P. and N.A.; formal analysis, K.P. and N.A.; behavior training, K.P., N.A., J.D.R., J.Z., K.C.J., M.L.-E., E.K., L.L., K.W., J.A., A.D.L., and Y.Y.; methodology, K.P. and B.H.; supervision, B.H.; validation, K.P. and N.A.; visualization, K.P. and B.H.; writing, K.P. and B.H.; funding acquisition, K.P., J.D.R., K.C.J., and B.H.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data

Pre-processed data to reproduce main results	This paper	doi:10.6084/m9.figshare.30375733

Experimental Models: Organisms/Strains

B6.Cg-Tg(Camk2a-tTA)1Mmay/DboJ	Jackson Laboratory	RRID:IMSR_JAX:007004
B6;DBA-Tg(tetO-GCaMP6s)2Niell/J	Jackson Laboratory	RRID:IMSR_JAX:024742
Ssttm2.1(cre)Zjh/J	Jackson Laboratory	RRID:IMSR_JAX:013044
B6;129S-Gt(ROSA)26Sortm32(CAG-COP4∗H134R/EYFP)Hze/J	Jackson Laboratory	RRID:IMSR_JAX:012569
B6.129P2-Pvalbtm1(cre)Arbr/J	Jackson Laboratory	RRID:IMSR_JAX:017320
B6;C3-Tg(Scnn1a-cre)3Aibs/J	Jackson Laboratory	RRID:IMSR_JAX:009613
C57BL/6J	Jackson Laboratory	RRID:IMSR_JAX:000664
B6.129S2-Emx1^tm1(cre)Krj/J	Jackson Laboratory	RRID:IMSR_JAX:005628

Software and algorithms

MATLAB 2024b or later	The Mathworks, Inc.	https://www.mathworks.com/products/matlab.html
Custom code to reproduce main results	This paper	doi:10.6084/m9.figshare.30375733

Open in a new tab

Experimental model and study participant details

All experiments were approved by the Institutional Animal Care and Use Committee at Georgia Institute of Technology and adhered to NIH guidelines. Male (N = 30) and female (N = 6) mice (8–12 weeks old, Jackson Labs) were used for all studies. Behavioral experiments were conducted in N = 5 Camk2a-tTA mice (RRID:IMSR_JAX:007004) crossed with tetO-GCaMP6s (RRID:IMSR_JAX:024742); N = 1 Sst-IRES-Cre (RRID: IMSR_JAX:013044) mice crossed with Ai32(RCL-ChR2(H134R)/EYFP) (RRID: IMSR_JAX:024109) mouse; N = 3 B6 PV[cre] (RRID: IMSR_JAX:017320) crossed with Ai32(RCL-ChR2(H134R)/EYFP) (RRID: IMSR_JAX:024109) mice; N = 9 Scnn1a-Tg3-Cre (RRID:IMSR_JAX:009613) crossed with Ai32(RCL-ChR2(H134R)/EYFP) (RRID: IMSR_JAX:024109) mice; N = 16 C57BL/6J (RRID:IMSR_JAX:000664) mice; and N = 2 Emx1-Cre (RRID:IMSR_JAX:005628) crossed with C57BL/6J (RRID:IMSR_JAX:000664) mice. There was no significant difference in total training time (up to Phase 3) across genotypes (p = 0.075, Kruskal–Wallis test). We generally used male mice more often than female mice simply because they are heavier even when age-matched, and thus complete more trials per session,⁵⁴ maximizing statistical power during a demanding behavioral task. Heavier male mice were also generally more robust to water restriction. We did not find sex differences for any attentional metrics (Table S1). Mice were bred in-house and group housed (2–4) in ventilated cages under a 12-h reverse light/dark cycle with temperature and humidity controlled (21°C–23°C, 40–60%) prior to being selected for experiments. Mice were randomly chosen for experiments, as long as they had no history of poor health, were the appropriate weight for their age, and met genetic requirements (i.e., were the correct genotype). When mice were selected for training, they were kept single housed as previously described.²⁰^,³³^,⁹⁰ Enrichment (e.g., nesting materials) was provided in all cages. Food and water were provided ad libitum before training; thereafter, water was removed from the cage and hydration followed the protocol as described in our prior studies²⁰^,³³^,⁹⁰ and below.

Method details

Headplate implantation

Mice were anesthetized with isoflurane (3% induction, 1–2% maintenance) and implanted with a custom stainless steel head plate with integrated recording chamber as described previously.²⁰^,⁹⁰ The head plate was attached to the skull using a layer of adhesive (VetBond) before being fully secured to the cranium using dental cement (Metabond). The recording chamber was sealed with an elastomer (KwikCast) to protect the skull. Following implantation, analgesics were administered immediately upon recovery from anesthesia (sustained-release buprenorphine, 1 mg/kg) and mice were given a minimum of three days to fully recover in their home cage.

Handling and habituation

Mice underwent gentle handling and habituation for at least three days before training. On day one, experimenters handled the mice in their home cages for 15–30 min. On day two, mice were handled for 30–45 min, and introduced to head-fixation in the home cage, but not placed on the training rig. By day three, the mice were head-fixed and placed on the training rig for 15 min, without any stimuli or the lick detector present. Each mouse was trained at the same time every day.

Water restriction and weight monitoring

The pre-restriction weight of each mouse was recorded as its baseline. Mice were single-housed and water-restricted for 24 h before the first behavioral task training day, and received an ad libitum mix of standard and high-fat diets to maintain weight (50/50 mix of Purina Lab Diet, Laboratory Rodent Diet 5001; Purina Lab Diet, Mouse Diet 5015). Weights were recorded both before and after training, along with the amount of water received during the task. Each correct trial provided 2.5–3.0 μL of water. The minimum amount of water given per day was calculated from the age-and strain-matched minimum expected weight from Jackson labs (typically 1.0mL/day for a 25g mouse). If the daily minimum water amount was not met during the task, mice were supplemented with Hydrogel, administered 60–120 min post-training. Hydrogel was administered daily, including weekends. If a mouse dropped below 75% of the minimum weight, as per prior studies⁹⁵ they were given the day off with full supplemental water and food (e.g., sunflower seeds, high-fat diet). If weight remained below the minimum for more than three days, the mouse was given ad libitum water and removed from training (1 of 39 mice in the current study).

Behavior apparatus

The mouse was head-fixed and rested comfortably on its hindlimbs in a semi-enclosed tube with forelimbs resting on a bar. The mouse was placed in the center of the behavior training apparatus on a custom-built platform (Figures S8A and S8B). Visual stimuli were displayed using MATLAB 2019b and Psychtoolbox version 3.0.19.⁹⁶^,⁹⁷ Two computer screens (gamma-corrected LCD displays; Dell Ultrasharp U2417H or U2419H; 60 Hz refresh rate; peak luminance of 250 cd ∗ m⁻²) were placed at 90° angles from one another. The two displays were positioned such that stimuli at 0° and 70° in azimuthal space have similar viewing angles relative to the central axis of the mouse’s right eye. Viewing distance from the mouse to each viewing screen was 21 cm. The training chamber was lined with sound-dampening material to minimize outside noise.

Stimuli presented during behavioral training were controlled using a customized version of Rigbox,⁹⁸ which logged stimulus presentation, licking events, reward delivery, and eye tracking through a National Instruments data acquisition (DAQ) board. These signals were acquired and logged at 20kHz. All task events (e.g., licks, reward delivery), stimulus parameters (e.g., stimulus duration, contrast, size, position) and behavioral responses (e.g., trial reaction times, trial behavioral outcome) were acquired for every behavioral training session, and timestamped and stored offline for further analysis.

A custom-built lick detector⁹⁹ was connected to a Tektronix oscilloscope (TBS1064) and licking events were observed in real time by experimenters. The lick detector stays at a fixed position for the duration of a training session. A water delivery spout was placed at the end of the lick detector, within reach of the animal’s tongue. When animals licked (either in response to the detection of a stimulus, or random licking) the tongue movement is picked up by a break in an infrared (IR) beam inside of the lick detector. The breaking of the IR beam during stimulus presentation triggered the release of a solenoid pinch valve that allowed for water to be delivered from the spout inside of the lick detector.

Eye camera

Face and eye movements were monitored by illuminating the animal’s face and eye with infrared light (Mightex, SLS-02008-A). The right eye was monitored with a camera (Imaging source DMK 21Bu04.H) fitted with a zoom lens (Navitar 7000) and a long-pass filter (Mightex, 092/52x0.75). The camera was placed ∼22 cm from the animal’s right eye. Videos were obtained at 30 Hz. These videos were monitored by the experimenter through the entire duration of training. Saved videos were analyzed for changes in pupil size and position using DeepLabCut.¹⁰⁰ Changes in pupil size are a well-established proxy for overall levels of arousal,¹⁰¹ but cannot distinguish between the sources of these potential changes (such as reward expectation or learning). In a subset of mice (11/36) we unfortunately discovered that one setup had corrupted video files during training and these could not be recovered for adequate analysis; the behavioral data in these mice was unaffected.

Task design

Throughout phases 1–4 (described below), mice were required to withhold licking for 0.5–15 s before stimulus onset. This mandatory quiescence period duration was randomized for each trial (exponential distribution). If the mouse licked during this period, the trial restarted with the same duration of enforced quiescence. Following stimulus onset, a lick during the response window (1s long) provided a water reward (hit trial) and turned off the stimulus. Failing to respond resulted in no reward (miss trial) and the next trial began. The stimulus is removed from the screen once the animal has responded, or after 1s (miss trials). A 0% contrast stimulus (‘blank’) was displayed on 20% of randomly selected trials. Licking during a ‘blank’ trial was recorded as a false alarm, and successfully withholding during a ‘blank’ trial was considered a correct rejection. Even though there was no punishment for licking during these trials, mice continually maintained relatively low false alarm rates during training (Figures 3C and 4G).

Definitions of trials, blocks, sessions, days

We define a trial as one single presentation of a stimulus (any contrast, including 0%); a block is defined as the collection of consecutive trials at a single stimulus position; a session is defined as the collection of blocks presented to the animal without any breaks (typically 100–200 trials total), with animals performing for 2–5 sessions per day. Within each session there are several block switches (3–7 per session) and each block consists of the presentation of the stimulus in one location for a given number of trials. Typically, at expert stage, each block consisted of 10 binocular trials and 20–30 monocular trials; trial numbers were unbalanced because animals tended to prioritize performing mostly binocular trials when both were equally available. A training day consisted of all trials across sessions combined. Animals typically performed 2–5 sessions per day, with 350–450 trials per day as experts in the task (369 ± 78, mean ± SD, N = 36). Experimenters controlled the start of successive sessions within a day. There was usually a pause lasting just a few minutes between sessions to save the data, assess performance metrics of the completed session, and adjust any parameters as needed for the next session.

Training Phase 0: Operant conditioning

Mice first learned to associate visual stimuli with water rewards through operant conditioning. High contrast (≥65%) Gabor gratings were presented in the center of the binocular region (0° azimuth). Mice were automatically rewarded with 2.5–3.0 μL water 0.7 s after stimulus onset, regardless of licking behavior. Reaction times were recorded, and once mice demonstrated a hit rate of ≥75% and a visual stimulus-triggered reaction time of ≤0.7 s (i.e., prior to reward onset) for at least 100–150 trials, automatic water delivery was discontinued, and reward was delivered only upon the first lick during the visual stimulus (operant conditioning). The response window typically spanned 2s after stimulus onset in Phase 0. With maintained hit rates and well-aligned visual reaction times (Figure S2), the response window was shortened to 1s (to incentivize fast reaction times) and mice progressed to Phase 1.

Training Phase 1: Eccentricity shifting

To train mice to detect stimuli in multiple spatial locations, the location of the Gabor grating was gradually shifted in horizontal (azimuth) space by 10°–20° increments starting from 0° (binocular region) and ending at ∼70° (monocular region). Mice were presented with a block of 50–100 consecutive trials at the same location. Once mice achieved ≥75% correct maintaining a d prime of >1, the stimulus was shifted by another 10°–20°. Mice typically displayed reaction times of 0.4–0.6 s. Once they maintained accuracy of ≥75% correct with a d prime >1 for a stimulus positioned at 70° for at least 1 training day (312 ± 70 trials performed per day in Phase 1; mean ± SD), they progressed to Phase 2.

Training Phase 2: Contrast shaping

In this phase, the stimulus remained positioned at ∼70°, but the contrast was varied trial-by-trial. We defined a set of logarithmically spaced contrasts (not including 0%), and four of these appeared during a given block. The range of contrasts progressively shifted to lower values according to performance, but drew from the set of defined reference contrasts. For example, contrast ranges start at [18, 33, 65, 85%] then decrease progressively from [10, 18, 33, 65%] to [5, 10, 18, 33%] to [2, 5, 10, 18%]. A given contrast range was presented in a block of at least 100 consecutive trials, and psychometric performance (hit rate across contrasts) was assessed after every session. The contrast ranges decreased in order to ensure psychometric performance spanned the full dynamic range of performance (from chance level hit rate at the lowest contrast to >75% hit rate at high contrast), with responses to mid-range contrasts falling between 25 and 75% accuracy. Once mice exhibited stable psychometric performance in the monocular region for 1–2 days (375 ± 108 trials performed per day in Phase 2; mean ± SD), they progressed to Phase 3.

Training Phase 3: Stimulus location switching

Stimuli now alternated between monocular (∼70°) and binocular (0°) positions in a blocked trial design to probe the behavioral effects of spatial attention. Blocks of 10–30 trials were presented consecutively at one location before switching to the other location without any warning or cueing. Performance metrics (reaction time, accuracy, psychometric performance) were analyzed within and across blocks to evaluate how detection speed and accuracy responded to changes in spatial context. By analyzing behavioral performance (hit rate, d’, reaction time) aligned to the block switches, we could assess if successive trials in the block showed improvements consistent with effects of spatial attention. Since sensitivity and performance were much better in binocular versus monocular visual space (Figures S3A and S3B), consistent with what we and others have shown previously, it was important to make sure the animal did not adopt a strategy of simply ignoring the monocular blocks, and waiting to only perform in the binocular blocks. Since monocular stimuli are inherently more difficult to detect, we present more of them for adequate reward rates and performance This was achieved by making the monocular blocks slightly longer (more rewarded trials) than the binocular blocks. The ratio of trials in [binocular: monocular] blocks typically started at [10:20], and was adjusted if needed per mouse based on psychometric performance in each location (average block sizes across N = 36 trained mice; binocular: 12 ± 2 trials; monocular: 22 ± 6 trials, mean ± SD). The most common adjustment was increasing the monocular block length if there was evidence of diminishing performance at that location. Animals were required to maintain psychometric performance with the highest contrast resulting in a d’ value >1 in both spatial locations to be considered experts. The full behavioral effects of spatial attention could be measured in Phase 3 (Figures S1C–S1E), and even on the first day of location switching in Phase 3. The effects did not get stronger, diminish, or suddenly emerge with practice opportunities in Phases 3 and 4 (See Video S1 for an example of a mouse performing the task during Phase 3). For animals that were trained for simultaneous neural recordings during behavior, we trained them additionally in Phase 4.

Video S1. Training in phase 3 during the task, related to STAR Methods

Download video file^{(3.8MB, mp4)}

Training Phase 4 (optional): Introduce task-irrelevant bars

We continued training exactly as described in Phase 3, but introduced low-contrast (±5%) task-irrelevant black and white vertical bars (10° wide, full screen height) presented one at a time (duration 0.1s or 0.4s) in random locations throughout the visual hemifield (spanning 160°). These bars were only used to measure neural receptive field responses during recordings, and to probe the influence of spatial attention on those responses. Our previous study showed that these low contrast bars did not interfere with detection of the higher contrast gratings.²⁰ Bars appeared during the entire behavioral session, during both inter-stimulus intervals and during stimulus events. Initially, bars are introduced at very low contrast (1–2% against gray background). If behavioral performance after the introduction of bars remained stable (d’ at highest contrast in both locations ≥1, psychometric performance in both locations), then bar contrasts slowly increased (in 1–2% increments) over training days until reaching the final 5% contrast value. See Video S2 for examples of mice during Phase 4. Phase 4 added an additional 9 ± 11 days of training, on average.

Video S2. Training in phase 4 during the task, related to STAR Methods

Download video file^{(871.5KB, mp4)}

Transitions between phases were determined by the experimenter, pending the guidance described above.

Subject inclusion/exclusion

A total of 39 mice were selected to be trained in the standardized protocol. Of these mice, 3/39 (8%) were excluded due to: headplate issues (1/39), falling below minimum weight during training (1/39), or failing to reach performance criteria to move on to the next phase (Phase 1 training, failed to reach >75% hit rate for high contrast monocular trials, 1/39). The remaining 36 mice (89%) satisfied inclusion criteria and completed the full task (Phase 0–3). Of these, 34/36 also completed the optional Phase 4. For Figure 4 (Contrast shaping), two mice were excluded from analysis because they only required contrast shaping for a single day, preventing early versus late comparisons across days.

Experimenter observations

Experimenters stayed in the same room as the mice for an entire training day. Experimenters trained only one mouse at any given time. During any training session, experimenters were not aware of what contrast or trial type was being shown to the animal on a given trial to minimize interference. Experimenters were able to observe licking activity, photodiode activity of visual stimulus events, and reward delivery in real-time on an oscilloscope. Videos of the face and eye were also recorded to observe pupil, face movements, and assess comfort, engagement, or stress levels of the mouse. After every session, experimenters combined observations of licking activity with quantitative behavioral output (Figures S8C–S8F) to adjust task parameters, if necessary.

Experimenter interventions

Mice typically performed a full behavioral session (50–200 trials) without experimenter intervention. The only real-time interventions experimenters made were to re-adjust the animal if sub-optimally positioned within the recording chamber, or adjusting the lick detector placement if observations indicated this was needed (e.g., the animal pushed it and reaction time latencies suddenly increased). These interventions were always made outside of the animal performing the task (i.e., the experimenter stopped the current stimulus set). These interventions typically occurred earlier in training and were rare later in training as mice gradually became more comfortable in the training environment (see Table S2). A session was stopped if the animal failed to respond for 7 trials in a row (regardless of trial type), or if the animal showed signs of discomfort. During such rest periods (∼5 min), single free rewards were given periodically to assess animal motivation (by examining lick rate evoked by free water reward consumption). Low lick rates indicated low motivation, while high lick rates (∼8 Hz) indicated the animal was still motivated by water reward and the session was restarted. If there was still lack of task engagement or continued signs of discomfort, animals were removed from the training apparatus and placed back in their home cage. Mice performed the task until they completely stopped licking (satiety) per above, and were then returned to the home cage.

Task parameters

Experimenters manually selected and adjusted the following task parameters according to the protocol: stimulus contrast, location, and the window of time for animals to withhold licking prior to stimulus onset (‘quiescence window’). The parameter choices were at the discretion of the trainer and determined by examining performance metrics in the previous days or sessions. Quiescence windows were drawn from an exponential distribution (flat hazard function), where experimenters defined the minimum, maximum, and median for the distribution. Each trial’s pre-stimulus quiescence window was drawn from this distribution. Longer quiescence windows were used to minimize false alarms (typical range: minimum 5s, maximum 15s, median 10s), while short windows were used when animals achieve expert-level performance in the task (typical range: minimum 1s, maximum 8s, median 5s). Between each session, these quiescence window parameters were randomly adjusted by 1–2 s to prevent animals from adopting a timing strategy. The average quiescence parameters for mice at the expert stage (during neural recordings) were: [minimum: 1.0s; maximum: 6.8s median: 4.5s].

Visual stimuli: Gabor gratings

The task-relevant rewarded stimuli were Gabor gratings with spatial frequency (0.05–0.1 cycles/degree) and size (σ = 10°) fixed across trials, while grating phase was randomized trial-to-trial (to mitigate adaptation). Stimuli were displayed against a 50% gray background.

Visual stimuli: Task-irrelevant bar stimuli

During the optional phase 4 of training, low-contrast (2–5%) black or white task-irrelevant vertical bars (10° wide, full screen height) were briefly presented (0.1 s duration, 0.3 s inter-stimulus interval in some cases; 0.4 s duration with no interval in other cases), one at a time, in a given position randomly from −45° to 117° (spanning 160° of the visual hemifield). See Video S3 for an example of a mouse performing the task after phase 4 training and during neural recordings. These vertical bars in phase 4 are the same as used in our previous studies for mapping spatiotemporal receptive fields in mouse primary visual cortex both during and outside of a behavioral task.²⁰^,³³^,⁹⁰^,¹⁰²

Video S3. Expert mouse performing the task, related to STAR Methods

Download video file^{(2.8MB, mp4)}

Performance metrics and behavioral analysis

Quantitative metrics used to assess task performance included hit rate, miss rate, false alarm rate, correct rejection rate, reaction time (RT), and psychometric curves to evaluate sensory thresholds. As in our prior studies,²⁰^,⁹⁰ we used signal detection theory to calculate sensitivity (d prime; d'). Behavioral analysis was automatically generated at the end of every session to track performance and inform experimenter decision making through training stages (Figures S8C–S8F). At the end of Phase 3, analysis was aligned to the first trial of block starts in the two stimulus locations (e.g., Figure 5) to assess effects of attentional shifts with location switching.

Quantification and statistical analysis

Quantification of attention effects

Behavioral evidence of attentional improvements included decreased reaction time, and/or increased accuracy and/or d’ on the later trials in the block following the stimulus location switch. To quantify this across the population, we used a linear mixed-effects model (fitlme, MATLAB 2024b) with trial number as a fixed effect and random intercepts for mouse and day, allowing us to assess attention-related changes both within and across animals (Figure 5).

To assess attention effects within individual training days during phases 3 and 4, we compared performance on each day between early trials (1–2) and late trials (6–7) for each mouse. For each day, four metrics were analyzed: reaction time (RT) and detection sensitivity (d’) for both monocular and binocular stimuli. To determine whether the observed effects exceeded what would be expected by chance, a permutation test was performed for every training day and metric across mice. Trial order was randomly shuffled 1000 times, and for each shuffle the difference between late and early trials was computed. The 95% confidence interval of this permutation-derived null distribution defined the range expected under the null hypothesis of ‘no attention effects.’ A day was labeled as a significant attention day if the observed late-early difference for any of the four metrics fell outside of this 95% confidence interval in the expected direction (e.g., decreases in the RT, increases in the d’).

In addition, to control for a potential confound of subject-to-subject differences in absolute reaction time range or d’ range driving the effects, we used a modulation index (MI) within subjects to calculate the paired relative change in reaction time and d prime between late (trials 6–7) and early (trials 1–2) trials within each day (Figure S1), defined as:

M o d u l a t i o n I n d e x (M I) = \frac{{L a t e}_{R T} - {E a r l y}_{R T}}{{L a t e}_{R T} + {E a r l y}_{R T}}

(Equation 1)

Because d prime values can be negative, the equation to calculate the MI for d prime changes across trials within a day was:

M o d u l a t i o n I n d e x (M I) = \frac{{L a t e}_{d^{'}} - {E a r l y}_{d^{'}}}{| {L a t e}_{d^{'}} | + | {E a r l y}_{d^{'}} |}

(Equation 2)

Perceptual contrast sensitivity: psychometric curve fitting

The hit rate (proportion of correct trials) was measured across contrasts. A Boltzmann’s sigmoidal equation was used to fit the data points to generate psychometric contrast response functions:

y = y_{\min} + \frac{y_{\max} - y_{\min}}{1 + e^{- \frac{x - x_{o}}{k}}}

(Equation 3)

y_min: minimum of the output variable (false alarm rate); y_max: maximum of the output variable (hit rate at highest contrast); x: contrast values; x_o: midpoint of the sigmoid curve (used to extract contrast threshold, c₅₀); k: steepness parameter. The data was fit using the MATLAB function ‘fit.’

Contrast sensitivity analysis

Psychometric sensitivity was analyzed across successive trials following a block switch. The label ‘monocular’ indicates all the trials following a switch from a binocular to monocular grating location. Likewise, the label ‘binocular’ indicates all the trials following a switch from a monocular to binocular grating location. In either case, behavioral data in phases 3 or 4 of training were aligned to the block switch, and hit trials aggregated across all block switches in all subjects. Since contrast levels for each trial were randomized (e.g., the first trial might display a stimulus at 10% contrast in one block and 18% contrast in another), all stimuli across all of the blocks can be used to construct a contrast responses curve from all of the first trials after the block switch. These aggregate psychometric curves were constructed for each successive trial and illustrate how the psychometric contrast sensitivity across all contrasts changes following the block switch (Figures 6A–6D). Contrasts were converted to proportions (0–1) prior to fitting and slope calculation. Contrast thresholds and slopes were derived from the psychometric curves, following a similar approach outlined in.³⁴

Psychometric curves were fit separately for monocular and binocular conditions using a four-parameter Boltzmann function (see above). Data were organized by animal, day, and trial number following a block switch. For some days, curve fits could not be computed for certain trial–contrast combinations due to insufficient trials (274 of 759 total days); these were excluded from further analysis. Only trials from days where sensitivity (d’) increased as a function of trial number within the block (monocular or binocular) were included, resulting in 234 of 485 total days. From each successfully computed fit, contrast threshold (c₅₀) and slope was calculated as:

s l o p e = \frac{y_{\max} - y_{\min}}{4 k}

(Equation 4)

Alternative slope estimations including calculating the slope of the line fit around ±5% of the data surrounding the c₅₀, and calculating the total change in hit rate over the total change in contrasts across the entire response curve without curve fitting yielded similar results. Linear mixed-effects models (fitlme, MATLAB 2024b) were used to assess changes in these parameters across trials after the block switch, including random intercepts for animal, and day nested within animal. A total of 36 animals contributed to this dataset, in which monocular data was collected from N = 34 animals (2 animals contributing binocular data only) and binocular data was collected from N = 35 animals (1 animal contributing monocular data only). Group averages (mean) and standard errors (SEM) were computed for visualization across successive trials after the block switch (Figure 6).

Psychophysical model

To characterize the computations underlying trial-by-trial improvements in contrast sensitivity, we constructed two psychophysical models based on established frameworks³⁶^,³⁷^,³⁹ of 1) contrast gain and 2) response gain. In both models, behavioral performance was described using a hyperbolic ratio (Naka-Rushton) function:

R = r_{\max} * \frac{c^{n}}{{c^{n} + c}_{50}^{n}}

(Equation 5)

R: response rate; r_max: maximum response rate; c: contrast value; c₅₀: contrast value generating 50% response rate; n: exponent (controlling slope).

Contrast gain model

In the contrast gain model (Figures S6B–S6D), the contrast threshold (c₅₀) was shifted leftward across successive trials while r_max and n were held constant. This reflects an increase in contrast sensitivity without a change in maximum response rate. For monocular conditions, where our tested contrast range was 2–65% in most animals, c₅₀ was decreased linearly from 0.16 to 0.08 across trial bins. For binocular conditions where our tested contrast range was narrower (2–33% in most animals), c₅₀ was decreased linearly from 0.06 to 0.03 across trial bins. The r_max was fixed at 1.00 and the exponent n was fixed at 2.5 for both monocular and binocular conditions.

Response gain model

In the response gain model (Figures S6E–S6G), performance improved via multiplicative increases in r_max across trial bins 1–5 while c₅₀ and n were held constant. This reflects an increase in overall response amplitude without a change in contrast sensitivity. The r_max increased across trial bins, starting at 0.65 and scaled by a factor of 1.10 per trial bin. c₅₀ was fixed at 0.10 and n was fixed at 2.5 for both monocular and binocular conditions.

c₅₀ and d’ calculation

To estimate trial-by-trial changes in performance, we simulated behavioral performance across trials. For each simulated day and trial bin, responses were generated across 10 block switches per day. On each switch, a contrast level was randomly selected from the stimulus set and a hit was generated based on the model-predicted hit rate at that contrast. False alarms were generated independently using a fixed false alarm rate of 0.20 (similar to our 0.23 false alarm rate). Hit rates and false alarm rates were pooled across switches within each trial bin. To avoid extreme values (e.g., a hit rate of 100%), rates were computed using a standard log-linear correction, and d’ was then calculated as the difference between the Z score of the hit rate and the Z score of the false alarm rate.²⁰^,¹⁰³ This procedure was repeated across 400 simulated days, and d’ was averaged across days for each trial bin.

For the model, c₅₀ corresponds to the contrast at which the response is at 50% of the maximum. c₅₀ was estimated as the contrast at which the modeled response reached half of r_max across trials, again repeated over the 400 simulated days.

Contrast response function analysis

To compare our data to each model, we computed the change in hit rate (HR) across trials 2–5, relative to trial 1 (ΔHR) as function of contrast, and compared this matrix [contrast x trial x ΔHR] for our real data (same dataset as in Figure 6, selecting for days with increasing d' across block switches) to those for each of the two models. Because individual animals were tested across variable contrast ranges (reflecting individual differences in contrast thresholds, daily contrast ranges), we used a binned approach to maximize the number of animals contributing data to each contrast bin. On any given day, an animal’s contrast set could vary slightly from other days. Contrast bins were therefore defined by fixed edges rather than singular values (monocular bins: 1–4%, 5–9%, 10–17%, 18–32%, 33–49%, ≥50%; binocular bins: 1–4%, 5–9%, 10–17%, 18–32%, ≥33%), and animals contributed data to a bin if they had at least one trial at a contrast falling within that range. Hit rates were averaged within each bin per animal per trial number after the block switch for all contributing days, and then averaged across animals. ΔHR was computed per animal before averaging across animals to control for between-animal differences in performance (Figures S6N–S6U).

Model comparison to dataset

Similarity between our data’s ΔHR matrices and each of the two models was quantified using the 2D correlation coefficient (corr2, MATLAB 2024b). Statistical significance was assessed via permutation testing, where the monocular and binocular ΔHR matrices were randomly shuffled (randomly permutating elements in both dimensions without replacement to remove the relationship between trial and contrast) 10,000 times and correlated with each model, generating a null distribution of correlation values. The p-value was defined as the proportion of permuted correlations exceeding the observed correlation.

To estimate the variability across the correlation values for each model, we performed bootstrap resampling across mice. For each iteration, mice were resampled with replacement (5k times), and ΔHR matrices were computed from the resampled data. Correlations between each of the two models (for both binocular and monocular data) were then calculated. 95% confidence intervals (CI) were defined as the 2.5^th and 97.5^th percentiles of the bootstrap distributions. A CI that spans 0 indicates weak correlations.

Visualization

All figures were generated using colorblind-friendly color palettes (magma, viridis, and Okabe-Ito) to improve accessibility for colorblind readers.

Statistical analysis

No specific methods were used to determine the need for replication, or strategy for randomization aside from the linear mixed effects model. No methods were used to determine statistical sample size. Data distribution for normality was not formally tested. Unless otherwise noted, signed rank tests for paired data were performed. Statistical significance was determined at α = 0.05. Neither experiments nor data analysis were performed blind to the conditions of the animal.

Published: July 7, 2026

Footnotes

Supplemental information can be found online at https://doi.org/10.1016/j.crmeth.2026.101514.

Supplemental information

Document S1. Figures S1–S8 and Tables S1–S3

mmc1.pdf^{(1.8MB, pdf)}

Document S2. Article plus supplemental information

mmc5.pdf^{(4.1MB, pdf)}

References

1.Aguillon-Rodriguez V., Angelaki D., Bayer H., Bonacchi N., Carandini M., Cazettes F., Chapuis G., Churchland A.K., Dan Y., Dewitt E., et al. Standardized and reproducible measurement of decision-making in mice. eLife. 2021;10 doi: 10.7554/eLife.63711. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Posner M.I., Snyder C.R., Davidson B.J. Attention and the detection of signals. J. Exp. Psychol. 1980;109:160–174. [PubMed] [Google Scholar]
3.Reynolds J.H., Chelazzi L. Attentional modulation of visual processing. Annu. Rev. Neurosci. 2004;27:611–647. doi: 10.1146/annurev.neuro.26.041002.131039. [DOI] [PubMed] [Google Scholar]
4.Carrasco M., McElree B. Covert attention accelerates the rate of visual information processing. Proc. Natl. Acad. Sci. USA. 2001;98:5363–5367. doi: 10.1073/pnas.081074098. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Maunsell J.H.R. Neuronal Mechanisms of Visual Attention. Annu. Rev. Vis. Sci. 2015;1:373–391. doi: 10.1146/annurev-vision-082114-035431. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Desimone R., Duncan J. Neural mechanisms of selective visual attention. Annu. Rev. Neurosci. 1995;18:193–222. doi: 10.1146/annurev.ne.18.030195.001205. [DOI] [PubMed] [Google Scholar]
7.Wurtz R.H., Mohler C.W. Enhancement of visual responses in monkey striate cortex and frontal eye fields. J. Neurophysiol. 1976;39:766–772. doi: 10.1152/jn.1976.39.4.766. [DOI] [PubMed] [Google Scholar]
8.McAlonan K., Cavanaugh J., Wurtz R.H. Attentional modulation of thalamic reticular neurons. J. Neurosci. 2006;26:4444–4450. doi: 10.1523/JNEUROSCI.5602-05.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Treue S. Neural correlates of attention in primate visual cortex. Trends Neurosci. 2001;24:295–300. doi: 10.1016/s0166-2236(00)01814-2. [DOI] [PubMed] [Google Scholar]
10.Talluri B.C., Kang I., Lazere A., Quinn K.R., Kaliss N., Yates J.L., Butts D.A., Nienborg H. Activity in primate visual cortex is minimally driven by spontaneous movements. Nat. Neurosci. 2023;26:1953–1959. doi: 10.1038/s41593-023-01459-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Wang L., Krauzlis R.J. Visual Selective Attention in Mice. Curr. Biol. 2018;28:676–685.e4. doi: 10.1016/j.cub.2018.01.038. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Gămănuţ R., Shimaoka D. Anatomical and functional connectomes underlying hierarchical visual processing in mouse visual system. Brain Struct. Funct. 2022;227:1297–1315. doi: 10.1007/s00429-021-02415-4. [DOI] [PubMed] [Google Scholar]
13.Krubitzer L. The magnificent compromise: cortical field evolution in mammals. Neuron. 2007;56:201–208. doi: 10.1016/j.neuron.2007.10.002. [DOI] [PubMed] [Google Scholar]
14.Niell C.M., Stryker M.P. Highly selective receptive fields in mouse visual cortex. J. Neurosci. 2008;28:7520–7536. doi: 10.1523/JNEUROSCI.0623-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Siegle J.H., Jia X., Durand S., Gale S., Bennett C., Graddis N., Heller G., Ramirez T.K., Choi H., Luviano J.A., et al. Survey of spiking in the mouse visual system reveals functional hierarchy. Nature. 2021;592:86–92. doi: 10.1038/s41586-020-03171-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Felleman D.J., Van Essen D.C. Distributed hierarchical processing in the primate cerebral cortex. Cereb. Cortex. 1991;1:1–47. doi: 10.1093/cercor/1.1.1-a. [DOI] [PubMed] [Google Scholar]
17.Hu F., Dan Y. An inferior-superior colliculus circuit controls auditory cue-directed visual spatial attention. Neuron. 2022;110:109–119.e3. doi: 10.1016/j.neuron.2021.10.004. [DOI] [PubMed] [Google Scholar]
18.McBride E.G., Lee S.Y.J., Callaway E.M. Local and Global Influences of Visual Spatial Selection and Locomotion in Mouse Primary Visual Cortex. Curr. Biol. 2019;29:1592–1605.e5. doi: 10.1016/j.cub.2019.03.065. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.You W.K., Mysore S.P. Endogenous and exogenous control of visuospatial selective attention in freely behaving mice. Nat. Commun. 2020;11:1986. doi: 10.1038/s41467-020-15909-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Speed A., Del Rosario J., Mikail N., Haider B. Spatial attention enhances network, cellular and subthreshold responses in mouse visual cortex. Nat. Commun. 2020;11:505. doi: 10.1038/s41467-020-14355-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Kanamori T., Mrsic-Flogel T.D. Independent response modulation of visual cortical neurons by attentional and behavioral states. Neuron. 2022;110:3907–3918.e6. doi: 10.1016/j.neuron.2022.08.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Lehnert J., Cha K., Halperin J., Yang K., Zheng D.F., Khadra A., Cook E.P., Krishnaswamy A. Visual attention to features and space in mice using reverse correlation. Curr. Biol. 2023;33:3690–3701.e4. doi: 10.1016/j.cub.2023.07.060. [DOI] [PubMed] [Google Scholar]
23.Norman K.J., Riceberg J.S., Koike H., Bateh J., McCraney S.E., Caro K., Kato D., Liang A., Yamamuro K., Flanigan M.E., et al. Post-error recruitment of frontal sensory cortical projections promotes attention in mice. Neuron. 2021;109:1202–1213.e5. doi: 10.1016/j.neuron.2021.02.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Steinmetz N.A., Koch C., Harris K.D., Carandini M. Challenges and opportunities for large-scale electrophysiology with Neuropixels probes. Curr. Opin. Neurobiol. 2018;50:92–100. doi: 10.1016/j.conb.2018.01.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Urai A.E., Doiron B., Leifer A.M., Churchland A.K. Large-scale neural recordings call for new insights to link brain and behavior. Nat. Neurosci. 2022;25:11–19. doi: 10.1038/s41593-021-00980-9. [DOI] [PubMed] [Google Scholar]
26.Guo Z.V., Hires S.A., Li N., O'Connor D.H., Komiyama T., Ophir E., Huber D., Bonardi C., Morandell K., Gutnisky D., et al. Procedures for behavioral experiments in head-fixed mice. PLoS One. 2014;9 doi: 10.1371/journal.pone.0088678. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Rosenberg M., Zhang T., Perona P., Meister M. Mice in a labyrinth show rapid learning, sudden insight, and efficient exploration. eLife. 2021;10 doi: 10.7554/eLife.66175. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Levy D.R., Hunter N., Lin S., Robinson E.M., Gillis W., Conlin E.B., Anyoha R., Shansky R.M., Datta S.R. Mouse spontaneous behavior reflects individual variation rather than estrous state. Curr. Biol. 2023;33:1358–1364.e4. doi: 10.1016/j.cub.2023.02.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Le N.M., Yildirim M., Wang Y., Sugihara H., Jazayeri M., Sur M. Mixtures of strategies underlie rodent behavior during reversal learning. PLoS Comput. Biol. 2023;19 doi: 10.1371/journal.pcbi.1011430. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Gilad A., Gallero-Salas Y., Groos D., Helmchen F. Behavioral Strategy Determines Frontal or Posterior Location of Short-Term Memory in Neocortex. Neuron. 2018;99:814–828.e7. doi: 10.1016/j.neuron.2018.07.029. [DOI] [PubMed] [Google Scholar]
31.van Beest E.H., Mukherjee S., Kirchberger L., Schnabel U.H., van der Togt C., Teeuwen R.R.M., Barsegyan A., Meyer A.F., Poort J., Roelfsema P.R., Self M.W. Mouse visual cortex contains a region of enhanced spatial resolution. Nat. Commun. 2021;12:4029. doi: 10.1038/s41467-021-24311-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Johnson K.P., Fitzpatrick M.J., Zhao L., Wang B., McCracken S., Williams P.R., Kerschensteiner D. Cell-type-specific binocular vision guides predation in mice. Neuron. 2021;109:1527–1539.e4. doi: 10.1016/j.neuron.2021.03.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Speed A., Del Rosario J., Burgess C.P., Haider B. Cortical State Fluctuations across Layers of V1 during Visual Spatial Perception. Cell Rep. 2019;26:2868–2874.e3. doi: 10.1016/j.celrep.2019.02.045. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Cohen M.R., Maunsell J.H.R. When attention wanders: how uncontrolled fluctuations in attention affect performance. J. Neurosci. 2011;31:15802–15806. doi: 10.1523/JNEUROSCI.3063-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Cameron E.L., Tai J.C., Carrasco M. Covert attention affects the psychometric function of contrast sensitivity. Vision Res. 2002;42:949–967. doi: 10.1016/s0042-6989(02)00039-1. [DOI] [PubMed] [Google Scholar]
36.Reynolds J.H., Heeger D.J. The normalization model of attention. Neuron. 2009;61:168–185. doi: 10.1016/j.neuron.2009.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Lee J., Maunsell J.H.R. A normalization model of attentional modulation of single unit responses. PLoS One. 2009;4 doi: 10.1371/journal.pone.0004651. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Reynolds J.H., Pasternak T., Desimone R. Attention increases sensitivity of V4 neurons. Neuron. 2000;26:703–714. doi: 10.1016/s0896-6273(00)81206-4. [DOI] [PubMed] [Google Scholar]
39.Herrmann K., Montaser-Kouhsari L., Carrasco M., Heeger D.J. When size matters: attention affects performance by contrast or response gain. Nat. Neurosci. 2010;13:1554–1559. doi: 10.1038/nn.2669. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Carrasco M. Visual attention: the past 25 years. Vision Res. 2011;51:1484–1525. doi: 10.1016/j.visres.2011.04.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Luo T.Z., Maunsell J.H.R. Attentional Changes in Either Criterion or Sensitivity Are Associated with Robust Modulations in Lateral Prefrontal Cortex. Neuron. 2018;97:1382–1393.e7. doi: 10.1016/j.neuron.2018.02.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Shah S., Mancarella M., Hembrook-Short J.R., Mock V.L., Briggs F. Attention differentially modulates multiunit activity in the lateral geniculate nucleus and V1 of macaque monkeys. J. Comp. Neurol. 2022;530:1064–1080. doi: 10.1002/cne.25168. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Krauzlis R.J., Nichols N., Rangarajan K.V., McAlonan K., Goldstein S., Yochelson D., Wang L. Visual Psychophysics in Head-Fixed Mice. Curr. Protoc. Neurosci. 2020;92 doi: 10.1002/cpns.95. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Neske G.T., Nestvogel D., Steffan P.J., McCormick D.A. Distinct Waking States for Strong Evoked Responses in Primary Visual Cortex and Optimal Visual Detection Performance. J. Neurosci. 2019;39:10044–10059. doi: 10.1523/JNEUROSCI.1226-18.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Wang L., Rangarajan K.V., Gerfen C.R., Krauzlis R.J. Activation of Striatal Neurons Causes a Perceptual Decision Bias during Visual Change Detection in Mice. Neuron. 2018;97:1369–1381.e5. doi: 10.1016/j.neuron.2018.01.049. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Niell C.M., Stryker M.P. Modulation of visual responses by behavioral state in mouse visual cortex. Neuron. 2010;65:472–479. doi: 10.1016/j.neuron.2010.01.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Mineault P.J., Tring E., Trachtenberg J.T., Ringach D.L. Enhanced Spatial Resolution During Locomotion and Heightened Attention in Mouse Primary Visual Cortex. J. Neurosci. 2016;36:6382–6392. doi: 10.1523/JNEUROSCI.0430-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Bennett C., Arroyo S., Hestrin S. Subthreshold mechanisms underlying state-dependent modulation of visual responses. Neuron. 2013;80:350–357. doi: 10.1016/j.neuron.2013.08.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Liska J.P., Rowley D.P., Nguyen T.T.K., Muthmann J.O., Butts D.A., Yates J., Huk A.C. Running modulates primate and rodent visual cortex differently. eLife. 2024;12 doi: 10.7554/eLife.87736. [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Cornsweet T.N. Academic Press; 1970. Visual Perception. [Google Scholar]
51.Hecht S., Shlaer S., Pirenne M.H. Energy, Quanta, and Vision. J. Gen. Physiol. 1942;25:819–840. doi: 10.1085/jgp.25.6.819. [DOI] [PMC free article] [PubMed] [Google Scholar]
52.van Vugt B., Dagnino B., Vartak D., Safaai H., Panzeri S., Dehaene S., Roelfsema P.R. The threshold for conscious report: Signal loss and response bias in visual and frontal cortex. Science. 2018;360:537–542. doi: 10.1126/science.aar7186. [DOI] [PubMed] [Google Scholar]
53.de Lafuente V., Romo R. Neuronal correlates of subjective sensory experience. Nat. Neurosci. 2005;8:1698–1703. doi: 10.1038/nn1587. [DOI] [PubMed] [Google Scholar]
54.Busse L., Ayaz A., Dhruv N.T., Katzner S., Saleem A.B., Schölvinck M.L., Zaharia A.D., Carandini M. The detection of visual contrast in the behaving mouse. J. Neurosci. 2011;31:11351–11361. doi: 10.1523/JNEUROSCI.6689-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Goldbach H.C., Akitake B., Leedy C.E., Histed M.H. Performance in even a simple perceptual task depends on mouse secondary visual areas. eLife. 2021;10 doi: 10.7554/eLife.62156. [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Cone J.J., Scantlen M.D., Histed M.H., Maunsell J.H.R. Different Inhibitory Interneuron Cell Classes Make Distinct Contributions to Visual Contrast Perception. eNeuro. 2019;6 doi: 10.1523/ENEURO.0337-18.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Dalgleish H.W., Russell L.E., Packer A.M., Roth A., Gauld O.M., Greenstreet F., Thompson E.J., Häusser M. How many neurons are sufficient for perception of cortical activity? eLife. 2020;9 doi: 10.7554/eLife.58889. [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Russell L.E., Fişek M., Yang Z., Tan L.P., Packer A.M., Dalgleish H.W.P., Chettih S.N., Harvey C.D., Häusser M. The influence of cortical activity on perception depends on behavioral state and sensory context. Nat. Commun. 2024;15:2456. doi: 10.1038/s41467-024-46484-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Maunsell J.H.R., Treue S. Feature-based attention in visual cortex. Trends Neurosci. 2006;29:317–322. doi: 10.1016/j.tins.2006.04.001. [DOI] [PubMed] [Google Scholar]
60.Luck S.J., Chelazzi L., Hillyard S.A., Desimone R. Neural mechanisms of spatial selective attention in areas V1, V2, and V4 of macaque visual cortex. J. Neurophysiol. 1997;77:24–42. doi: 10.1152/jn.1997.77.1.24. [DOI] [PubMed] [Google Scholar]
61.Ito M., Gilbert C.D. Attention modulates contextual influences in the primary visual cortex of alert monkeys. Neuron. 1999;22:593–604. doi: 10.1016/s0896-6273(00)80713-8. [DOI] [PubMed] [Google Scholar]
62.Maunsell J.H.R. Neuronal representations of cognitive state: reward or attention? Trends Cogn. Sci. 2004;8:261–265. doi: 10.1016/j.tics.2004.04.003. [DOI] [PubMed] [Google Scholar]
63.Krauzlis R.J., Wang L., Yu G., Katz L.N. What is attention? WIREs Cognit. Sci. 2023;14 doi: 10.1002/wcs.1570. [DOI] [PubMed] [Google Scholar]
64.Findling C., Hubert F., Acerbi L., Benson B., Benson J., Birman D., Bonacchi N., Buchanan E.K., Bruijns S., Carandini M., et al. Brain-wide representations of prior information in mouse decision-making. Nature. 2025;645:192–200. doi: 10.1038/s41586-025-09226-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Majumdar A., Ashcroft C., Fritsche M., Tan S., Zatka-Haas P., Folsz O., Walker N., Mistry L., Rominto A.M., Tvrdic M., et al. Distinct representations of economic variables across regions and projections of the frontal cortex. Neuron. 2025;113:4232–4244.e8. doi: 10.1016/j.neuron.2025.09.027. [DOI] [PubMed] [Google Scholar]
66.Fritsche M., Majumdar A., Strickland L., Liebana Garcia S., Bogacz R., Lak A. Temporal regularities shape perceptual decisions and striatal dopamine signals. Nat. Commun. 2024;15:7093. doi: 10.1038/s41467-024-51393-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Baruni J.K., Lau B., Salzman C.D. Reward expectation differentially modulates attentional behavior and activity in visual area V4. Nat. Neurosci. 2015;18:1656–1663. doi: 10.1038/nn.4141. [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Geng J.J., Behrmann M. Spatial probability as an attentional cue in visual search. Percept. Psychophys. 2005;67:1252–1268. doi: 10.3758/bf03193557. [DOI] [PubMed] [Google Scholar]
69.Druker M., Anderson B. Spatial probability AIDS visual stimulus discrimination. Front. Hum. Neurosci. 2010;4 doi: 10.3389/fnhum.2010.00063. [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Schnabel U.H., Van der Bijl T., Roelfsema P.R., Lorteije J.A.M. A Direct Comparison of Spatial Attention and Stimulus-Response Compatibility between Mice and Humans. J. Cogn. Neurosci. 2021;33:771–783. doi: 10.1162/jocn_a_01681. [DOI] [PubMed] [Google Scholar]
71.Cohen M.R., Maunsell J.H.R. Attention improves performance primarily by reducing interneuronal correlations. Nat. Neurosci. 2009;12:1594–1600. doi: 10.1038/nn.2439. [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Snyder A.C., Yu B.M., Smith M.A. Distinct population codes for attention in the absence and presence of visual stimulation. Nat. Commun. 2018;9:4382. doi: 10.1038/s41467-018-06754-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
73.McAdams C.J., Maunsell J.H.R. Attention to both space and feature modulates neuronal responses in macaque area V4. J. Neurophysiol. 2000;83:1751–1755. doi: 10.1152/jn.2000.83.3.1751. [DOI] [PubMed] [Google Scholar]
74.Connor C.E., Preddie D.C., Gallant J.L., Van Essen D.C. Spatial attention effects in macaque area V4. J. Neurosci. 1997;17:3201–3214. doi: 10.1523/JNEUROSCI.17-09-03201.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
75.McAdams C.J., Reid R.C. Attention modulates the responses of simple cells in monkey primary visual cortex. J. Neurosci. 2005;25:11023–11033. doi: 10.1523/JNEUROSCI.2904-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
76.Steinmetz N.A., Buetfering C., Lecoq J., Lee C.R., Peters A.J., Jacobs E.A.K., Coen P., Ollerenshaw D.R., Valley M.T., de Vries S.E.J., et al. Aberrant Cortical Activity in Multiple GCaMP6-Expressing Transgenic Mouse Lines. eNeuro. 2017;4 doi: 10.1523/ENEURO.0207-17.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
77.Loos M., Koopmans B., Aarts E., Maroteaux G., van der Sluis S., Verhage M., Smit A.B., Smit A.B. Within-strain variation in behavior differs consistently between common inbred strains of mice. Mamm. Genome. 2015;26:348–354. doi: 10.1007/s00335-015-9578-7. [DOI] [PubMed] [Google Scholar]
78.Jaric I., Voelkl B., Amrein I., Wolfer D.P., Novak J., Detotto C., Weber-Stadlbauer U., Meyer U., Manuella F., Mansuy I.M., Würbel H. Using mice from different breeding sites fails to improve replicability of results from single-laboratory studies. Lab Anim (NY) 2024;53:18–22. doi: 10.1038/s41684-023-01307-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
79.Voelkl B., Altman N.S., Forsman A., Forstmeier W., Gurevitch J., Jaric I., Karp N.A., Kas M.J., Schielzeth H., Van de Casteele T., Würbel H. Reproducibility of animal research in light of biological variation. Nat. Rev. Neurosci. 2020;21:384–393. doi: 10.1038/s41583-020-0313-3. [DOI] [PubMed] [Google Scholar]
80.Francis N.A., Bohlke K., Kanold P.O. Automated Behavioral Experiments in Mice Reveal Periodic Cycles of Task Engagement within Circadian Rhythms. eNeuro. 2019;6 doi: 10.1523/ENEURO.0121-19.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
81.Tsao C.H., Flint J., Huang G.J. Influence of diurnal phase on behavioral tests of sensorimotor performance, anxiety, learning and memory in mice. Sci. Rep. 2022;12:432. doi: 10.1038/s41598-021-03155-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
82.Georgiou P., Zanos P., Mou T.C.M., An X., Gerhard D.M., Dryanovski D.I., Potter L.E., Highland J.N., Jenne C.E., Stewart B.W., et al. Experimenters' sex modulates mouse behaviors and neural responses to ketamine via corticotropin releasing factor. Nat. Neurosci. 2022;25:1191–1200. doi: 10.1038/s41593-022-01146-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
83.Goldstein S., Wang L., McAlonan K., Torres-Cruz M., Krauzlis R.J. Stimulus-driven visual attention in mice. J. Vis. 2022;22:11. doi: 10.1167/jov.22.1.11. [DOI] [PMC free article] [PubMed] [Google Scholar]
84.Funayama K., Hagura N., Ban H., Ikegaya Y. Functional Organization of Flash-Induced V1 Offline Reactivation. J. Neurosci. 2016;36:11727–11738. doi: 10.1523/JNEUROSCI.1575-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
85.Funayama K., Minamisawa G., Matsumoto N., Ban H., Chan A.W., Matsuki N., Murphy T.H., Ikegaya Y. Neocortical Rebound Depolarization Enhances Visual Perception. PLoS Biol. 2015;13 doi: 10.1371/journal.pbio.1002231. [DOI] [PMC free article] [PubMed] [Google Scholar]
86.Jin M., Beck J.M., Glickfeld L.L. Neuronal Adaptation Reveals a Suboptimal Decoding of Orientation Tuned Populations in the Mouse Visual Cortex. J. Neurosci. 2019;39:3867–3881. doi: 10.1523/JNEUROSCI.3172-18.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
87.Chandrasekaran A.N., Vermani A., Gupta P., Steinmetz N., Moore T., Sridharan D. Dissociable components of attention exhibit distinct neuronal signatures in primate visual cortex. Sci. Adv. 2024;10 doi: 10.1126/sciadv.adi0645. [DOI] [PMC free article] [PubMed] [Google Scholar]
88.Carandini M., Churchland A.K. Probing perceptual decisions in rodents. Nat. Neurosci. 2013;16:824–831. doi: 10.1038/nn.3410. [DOI] [PMC free article] [PubMed] [Google Scholar]
89.Steinmetz N.A., Zatka-Haas P., Carandini M., Harris K.D. Distributed coding of choice, action and engagement across the mouse brain. Nature. 2019;576:266–273. doi: 10.1038/s41586-019-1787-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
90.Del Rosario J., Coletta S., Kim S.H., Mobille Z., Peelman K., Williams B., Otsuki A.J., Del Castillo Valerio A., Worden K., Blanpain L.T., et al. Lateral inhibition in V1 controls neural and perceptual contrast sensitivity. Nat. Neurosci. 2025;28:836–847. doi: 10.1038/s41593-025-01888-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
91.Glickfeld L.L., Histed M.H., Maunsell J.H.R. Mouse primary visual cortex is used to detect both orientation and contrast changes. J. Neurosci. 2013;33:19416–19422. doi: 10.1523/JNEUROSCI.3560-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
92.Wolff S.B., Ölveczky B.P. The promise and perils of causal circuit manipulations. Curr. Opin. Neurobiol. 2018;49:84–94. doi: 10.1016/j.conb.2018.01.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
93.Ross L.N., Bassett D.S. Causation in neuroscience: keeping mechanism meaningful. Nat. Rev. Neurosci. 2024;25:81–90. doi: 10.1038/s41583-023-00778-7. [DOI] [PubMed] [Google Scholar]
94.Ruediger S., Scanziani M. Learning speed and detection sensitivity controlled by distinct cortico-fugal neurons in visual cortex. eLife. 2020;9 doi: 10.7554/eLife.59247. [DOI] [PMC free article] [PubMed] [Google Scholar]
95.Burgess C.P., Lak A., Steinmetz N.A., Zatka-Haas P., Bai Reddy C., Jacobs E.A.K., Linden J.F., Paton J.J., Ranson A., Schröder S., et al. High-Yield Methods for Accurate Two-Alternative Visual Psychophysics in Head-Fixed Mice. Cell Rep. 2017;20:2513–2524. doi: 10.1016/j.celrep.2017.08.047. [DOI] [PMC free article] [PubMed] [Google Scholar]
96.Brainard D.H. The Psychophysics Toolbox. Spat. Vis. 1997;10:433–436. [PubMed] [Google Scholar]
97.Pelli D.G. The VideoToolbox software for visual psychophysics: transforming numbers into movies. Spat. Vis. 1997;10:437–442. [PubMed] [Google Scholar]
98.Bhagat J., Wells M.J., Harris K.D., Carandini M., Burgess C.P. Rigbox: An Open-Source Toolbox for Probing Neurons and Behavior. eNeuro. 2020;7 doi: 10.1523/ENEURO.0406-19.2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
99.Williams B., Speed A., Haider B. A novel device for real-time measurement and manipulation of licking behavior in head-fixed mice. J. Neurophysiol. 2018;120:2975–2987. doi: 10.1152/jn.00500.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
100.Mathis A., Mamidanna P., Cury K.M., Abe T., Murthy V.N., Mathis M.W., Bethge M. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat. Neurosci. 2018;21:1281–1289. doi: 10.1038/s41593-018-0209-y. [DOI] [PubMed] [Google Scholar]
101.McGinley M.J., Vinck M., Reimer J., Batista-Brito R., Zagha E., Cadwell C.R., Tolias A.S., Cardin J.A., McCormick D.A. Waking State: Rapid Variations Modulate Neural and Behavioral Responses. Neuron. 2015;87:1143–1161. doi: 10.1016/j.neuron.2015.09.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
102.Williams B., Del Rosario J., Muzzu T., Peelman K., Coletta S., Bichler E.K., Speed A., Meyer-Baese L., Saleem A.B., Haider B. Spatial modulation of dark versus bright stimulus responses in the mouse visual system. Curr. Biol. 2021;31:4172–4179.e6. doi: 10.1016/j.cub.2021.06.094. [DOI] [PMC free article] [PubMed] [Google Scholar]
103.Green D.M., Swets J.A. Wiley; 1966. Signal Detection Theory and Psychophysics. [Google Scholar]

Associated Data

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

Supplementary Materials

Video S1. Training in phase 3 during the task, related to STAR Methods

Download video file^{(3.8MB, mp4)}

Video S2. Training in phase 4 during the task, related to STAR Methods

Download video file^{(871.5KB, mp4)}

Video S3. Expert mouse performing the task, related to STAR Methods

Download video file^{(2.8MB, mp4)}

Document S1. Figures S1–S8 and Tables S1–S3

mmc1.pdf^{(1.8MB, pdf)}

Document S2. Article plus supplemental information

mmc5.pdf^{(4.1MB, pdf)}

Data Availability Statement

•
All data structures used to generate each figure are publicly available as of the date of publication at DOI: 10.6084/m9.figshare.30375733 and linked from the corresponding author’s institutional webpage.
•
All original code used to generate each figure are publicly available as of the date of publication at DOI: 10.6084/m9.figshare.30375733 and linked from the corresponding author’s institutional webpage.
•
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

[bib1] 1.Aguillon-Rodriguez V., Angelaki D., Bayer H., Bonacchi N., Carandini M., Cazettes F., Chapuis G., Churchland A.K., Dan Y., Dewitt E., et al. Standardized and reproducible measurement of decision-making in mice. eLife. 2021;10 doi: 10.7554/eLife.63711. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib2] 2.Posner M.I., Snyder C.R., Davidson B.J. Attention and the detection of signals. J. Exp. Psychol. 1980;109:160–174. [PubMed] [Google Scholar]

[bib3] 3.Reynolds J.H., Chelazzi L. Attentional modulation of visual processing. Annu. Rev. Neurosci. 2004;27:611–647. doi: 10.1146/annurev.neuro.26.041002.131039. [DOI] [PubMed] [Google Scholar]

[bib4] 4.Carrasco M., McElree B. Covert attention accelerates the rate of visual information processing. Proc. Natl. Acad. Sci. USA. 2001;98:5363–5367. doi: 10.1073/pnas.081074098. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib5] 5.Maunsell J.H.R. Neuronal Mechanisms of Visual Attention. Annu. Rev. Vis. Sci. 2015;1:373–391. doi: 10.1146/annurev-vision-082114-035431. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib6] 6.Desimone R., Duncan J. Neural mechanisms of selective visual attention. Annu. Rev. Neurosci. 1995;18:193–222. doi: 10.1146/annurev.ne.18.030195.001205. [DOI] [PubMed] [Google Scholar]

[bib7] 7.Wurtz R.H., Mohler C.W. Enhancement of visual responses in monkey striate cortex and frontal eye fields. J. Neurophysiol. 1976;39:766–772. doi: 10.1152/jn.1976.39.4.766. [DOI] [PubMed] [Google Scholar]

[bib8] 8.McAlonan K., Cavanaugh J., Wurtz R.H. Attentional modulation of thalamic reticular neurons. J. Neurosci. 2006;26:4444–4450. doi: 10.1523/JNEUROSCI.5602-05.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] 9.Treue S. Neural correlates of attention in primate visual cortex. Trends Neurosci. 2001;24:295–300. doi: 10.1016/s0166-2236(00)01814-2. [DOI] [PubMed] [Google Scholar]

[bib10] 10.Talluri B.C., Kang I., Lazere A., Quinn K.R., Kaliss N., Yates J.L., Butts D.A., Nienborg H. Activity in primate visual cortex is minimally driven by spontaneous movements. Nat. Neurosci. 2023;26:1953–1959. doi: 10.1038/s41593-023-01459-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] 11.Wang L., Krauzlis R.J. Visual Selective Attention in Mice. Curr. Biol. 2018;28:676–685.e4. doi: 10.1016/j.cub.2018.01.038. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib12] 12.Gămănuţ R., Shimaoka D. Anatomical and functional connectomes underlying hierarchical visual processing in mouse visual system. Brain Struct. Funct. 2022;227:1297–1315. doi: 10.1007/s00429-021-02415-4. [DOI] [PubMed] [Google Scholar]

[bib13] 13.Krubitzer L. The magnificent compromise: cortical field evolution in mammals. Neuron. 2007;56:201–208. doi: 10.1016/j.neuron.2007.10.002. [DOI] [PubMed] [Google Scholar]

[bib14] 14.Niell C.M., Stryker M.P. Highly selective receptive fields in mouse visual cortex. J. Neurosci. 2008;28:7520–7536. doi: 10.1523/JNEUROSCI.0623-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] 15.Siegle J.H., Jia X., Durand S., Gale S., Bennett C., Graddis N., Heller G., Ramirez T.K., Choi H., Luviano J.A., et al. Survey of spiking in the mouse visual system reveals functional hierarchy. Nature. 2021;592:86–92. doi: 10.1038/s41586-020-03171-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] 16.Felleman D.J., Van Essen D.C. Distributed hierarchical processing in the primate cerebral cortex. Cereb. Cortex. 1991;1:1–47. doi: 10.1093/cercor/1.1.1-a. [DOI] [PubMed] [Google Scholar]

[bib17] 17.Hu F., Dan Y. An inferior-superior colliculus circuit controls auditory cue-directed visual spatial attention. Neuron. 2022;110:109–119.e3. doi: 10.1016/j.neuron.2021.10.004. [DOI] [PubMed] [Google Scholar]

[bib18] 18.McBride E.G., Lee S.Y.J., Callaway E.M. Local and Global Influences of Visual Spatial Selection and Locomotion in Mouse Primary Visual Cortex. Curr. Biol. 2019;29:1592–1605.e5. doi: 10.1016/j.cub.2019.03.065. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] 19.You W.K., Mysore S.P. Endogenous and exogenous control of visuospatial selective attention in freely behaving mice. Nat. Commun. 2020;11:1986. doi: 10.1038/s41467-020-15909-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] 20.Speed A., Del Rosario J., Mikail N., Haider B. Spatial attention enhances network, cellular and subthreshold responses in mouse visual cortex. Nat. Commun. 2020;11:505. doi: 10.1038/s41467-020-14355-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib21] 21.Kanamori T., Mrsic-Flogel T.D. Independent response modulation of visual cortical neurons by attentional and behavioral states. Neuron. 2022;110:3907–3918.e6. doi: 10.1016/j.neuron.2022.08.028. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib22] 22.Lehnert J., Cha K., Halperin J., Yang K., Zheng D.F., Khadra A., Cook E.P., Krishnaswamy A. Visual attention to features and space in mice using reverse correlation. Curr. Biol. 2023;33:3690–3701.e4. doi: 10.1016/j.cub.2023.07.060. [DOI] [PubMed] [Google Scholar]

[bib23] 23.Norman K.J., Riceberg J.S., Koike H., Bateh J., McCraney S.E., Caro K., Kato D., Liang A., Yamamuro K., Flanigan M.E., et al. Post-error recruitment of frontal sensory cortical projections promotes attention in mice. Neuron. 2021;109:1202–1213.e5. doi: 10.1016/j.neuron.2021.02.001. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib24] 24.Steinmetz N.A., Koch C., Harris K.D., Carandini M. Challenges and opportunities for large-scale electrophysiology with Neuropixels probes. Curr. Opin. Neurobiol. 2018;50:92–100. doi: 10.1016/j.conb.2018.01.009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib25] 25.Urai A.E., Doiron B., Leifer A.M., Churchland A.K. Large-scale neural recordings call for new insights to link brain and behavior. Nat. Neurosci. 2022;25:11–19. doi: 10.1038/s41593-021-00980-9. [DOI] [PubMed] [Google Scholar]

[bib26] 26.Guo Z.V., Hires S.A., Li N., O'Connor D.H., Komiyama T., Ophir E., Huber D., Bonardi C., Morandell K., Gutnisky D., et al. Procedures for behavioral experiments in head-fixed mice. PLoS One. 2014;9 doi: 10.1371/journal.pone.0088678. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] 27.Rosenberg M., Zhang T., Perona P., Meister M. Mice in a labyrinth show rapid learning, sudden insight, and efficient exploration. eLife. 2021;10 doi: 10.7554/eLife.66175. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] 28.Levy D.R., Hunter N., Lin S., Robinson E.M., Gillis W., Conlin E.B., Anyoha R., Shansky R.M., Datta S.R. Mouse spontaneous behavior reflects individual variation rather than estrous state. Curr. Biol. 2023;33:1358–1364.e4. doi: 10.1016/j.cub.2023.02.035. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] 29.Le N.M., Yildirim M., Wang Y., Sugihara H., Jazayeri M., Sur M. Mixtures of strategies underlie rodent behavior during reversal learning. PLoS Comput. Biol. 2023;19 doi: 10.1371/journal.pcbi.1011430. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib30] 30.Gilad A., Gallero-Salas Y., Groos D., Helmchen F. Behavioral Strategy Determines Frontal or Posterior Location of Short-Term Memory in Neocortex. Neuron. 2018;99:814–828.e7. doi: 10.1016/j.neuron.2018.07.029. [DOI] [PubMed] [Google Scholar]

[bib31] 31.van Beest E.H., Mukherjee S., Kirchberger L., Schnabel U.H., van der Togt C., Teeuwen R.R.M., Barsegyan A., Meyer A.F., Poort J., Roelfsema P.R., Self M.W. Mouse visual cortex contains a region of enhanced spatial resolution. Nat. Commun. 2021;12:4029. doi: 10.1038/s41467-021-24311-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib32] 32.Johnson K.P., Fitzpatrick M.J., Zhao L., Wang B., McCracken S., Williams P.R., Kerschensteiner D. Cell-type-specific binocular vision guides predation in mice. Neuron. 2021;109:1527–1539.e4. doi: 10.1016/j.neuron.2021.03.010. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib33] 33.Speed A., Del Rosario J., Burgess C.P., Haider B. Cortical State Fluctuations across Layers of V1 during Visual Spatial Perception. Cell Rep. 2019;26:2868–2874.e3. doi: 10.1016/j.celrep.2019.02.045. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib34] 34.Cohen M.R., Maunsell J.H.R. When attention wanders: how uncontrolled fluctuations in attention affect performance. J. Neurosci. 2011;31:15802–15806. doi: 10.1523/JNEUROSCI.3063-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib35] 35.Cameron E.L., Tai J.C., Carrasco M. Covert attention affects the psychometric function of contrast sensitivity. Vision Res. 2002;42:949–967. doi: 10.1016/s0042-6989(02)00039-1. [DOI] [PubMed] [Google Scholar]

[bib36] 36.Reynolds J.H., Heeger D.J. The normalization model of attention. Neuron. 2009;61:168–185. doi: 10.1016/j.neuron.2009.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib37] 37.Lee J., Maunsell J.H.R. A normalization model of attentional modulation of single unit responses. PLoS One. 2009;4 doi: 10.1371/journal.pone.0004651. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib38] 38.Reynolds J.H., Pasternak T., Desimone R. Attention increases sensitivity of V4 neurons. Neuron. 2000;26:703–714. doi: 10.1016/s0896-6273(00)81206-4. [DOI] [PubMed] [Google Scholar]

[bib39] 39.Herrmann K., Montaser-Kouhsari L., Carrasco M., Heeger D.J. When size matters: attention affects performance by contrast or response gain. Nat. Neurosci. 2010;13:1554–1559. doi: 10.1038/nn.2669. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib40] 40.Carrasco M. Visual attention: the past 25 years. Vision Res. 2011;51:1484–1525. doi: 10.1016/j.visres.2011.04.012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib41] 41.Luo T.Z., Maunsell J.H.R. Attentional Changes in Either Criterion or Sensitivity Are Associated with Robust Modulations in Lateral Prefrontal Cortex. Neuron. 2018;97:1382–1393.e7. doi: 10.1016/j.neuron.2018.02.007. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] 42.Shah S., Mancarella M., Hembrook-Short J.R., Mock V.L., Briggs F. Attention differentially modulates multiunit activity in the lateral geniculate nucleus and V1 of macaque monkeys. J. Comp. Neurol. 2022;530:1064–1080. doi: 10.1002/cne.25168. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib43] 43.Krauzlis R.J., Nichols N., Rangarajan K.V., McAlonan K., Goldstein S., Yochelson D., Wang L. Visual Psychophysics in Head-Fixed Mice. Curr. Protoc. Neurosci. 2020;92 doi: 10.1002/cpns.95. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib44] 44.Neske G.T., Nestvogel D., Steffan P.J., McCormick D.A. Distinct Waking States for Strong Evoked Responses in Primary Visual Cortex and Optimal Visual Detection Performance. J. Neurosci. 2019;39:10044–10059. doi: 10.1523/JNEUROSCI.1226-18.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib45] 45.Wang L., Rangarajan K.V., Gerfen C.R., Krauzlis R.J. Activation of Striatal Neurons Causes a Perceptual Decision Bias during Visual Change Detection in Mice. Neuron. 2018;97:1369–1381.e5. doi: 10.1016/j.neuron.2018.01.049. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib46] 46.Niell C.M., Stryker M.P. Modulation of visual responses by behavioral state in mouse visual cortex. Neuron. 2010;65:472–479. doi: 10.1016/j.neuron.2010.01.033. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib47] 47.Mineault P.J., Tring E., Trachtenberg J.T., Ringach D.L. Enhanced Spatial Resolution During Locomotion and Heightened Attention in Mouse Primary Visual Cortex. J. Neurosci. 2016;36:6382–6392. doi: 10.1523/JNEUROSCI.0430-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib48] 48.Bennett C., Arroyo S., Hestrin S. Subthreshold mechanisms underlying state-dependent modulation of visual responses. Neuron. 2013;80:350–357. doi: 10.1016/j.neuron.2013.08.007. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib49] 49.Liska J.P., Rowley D.P., Nguyen T.T.K., Muthmann J.O., Butts D.A., Yates J., Huk A.C. Running modulates primate and rodent visual cortex differently. eLife. 2024;12 doi: 10.7554/eLife.87736. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib50] 50.Cornsweet T.N. Academic Press; 1970. Visual Perception. [Google Scholar]

[bib51] 51.Hecht S., Shlaer S., Pirenne M.H. Energy, Quanta, and Vision. J. Gen. Physiol. 1942;25:819–840. doi: 10.1085/jgp.25.6.819. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib52] 52.van Vugt B., Dagnino B., Vartak D., Safaai H., Panzeri S., Dehaene S., Roelfsema P.R. The threshold for conscious report: Signal loss and response bias in visual and frontal cortex. Science. 2018;360:537–542. doi: 10.1126/science.aar7186. [DOI] [PubMed] [Google Scholar]

[bib53] 53.de Lafuente V., Romo R. Neuronal correlates of subjective sensory experience. Nat. Neurosci. 2005;8:1698–1703. doi: 10.1038/nn1587. [DOI] [PubMed] [Google Scholar]

[bib54] 54.Busse L., Ayaz A., Dhruv N.T., Katzner S., Saleem A.B., Schölvinck M.L., Zaharia A.D., Carandini M. The detection of visual contrast in the behaving mouse. J. Neurosci. 2011;31:11351–11361. doi: 10.1523/JNEUROSCI.6689-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib55] 55.Goldbach H.C., Akitake B., Leedy C.E., Histed M.H. Performance in even a simple perceptual task depends on mouse secondary visual areas. eLife. 2021;10 doi: 10.7554/eLife.62156. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib56] 56.Cone J.J., Scantlen M.D., Histed M.H., Maunsell J.H.R. Different Inhibitory Interneuron Cell Classes Make Distinct Contributions to Visual Contrast Perception. eNeuro. 2019;6 doi: 10.1523/ENEURO.0337-18.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib57] 57.Dalgleish H.W., Russell L.E., Packer A.M., Roth A., Gauld O.M., Greenstreet F., Thompson E.J., Häusser M. How many neurons are sufficient for perception of cortical activity? eLife. 2020;9 doi: 10.7554/eLife.58889. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib58] 58.Russell L.E., Fişek M., Yang Z., Tan L.P., Packer A.M., Dalgleish H.W.P., Chettih S.N., Harvey C.D., Häusser M. The influence of cortical activity on perception depends on behavioral state and sensory context. Nat. Commun. 2024;15:2456. doi: 10.1038/s41467-024-46484-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib59] 59.Maunsell J.H.R., Treue S. Feature-based attention in visual cortex. Trends Neurosci. 2006;29:317–322. doi: 10.1016/j.tins.2006.04.001. [DOI] [PubMed] [Google Scholar]

[bib60] 60.Luck S.J., Chelazzi L., Hillyard S.A., Desimone R. Neural mechanisms of spatial selective attention in areas V1, V2, and V4 of macaque visual cortex. J. Neurophysiol. 1997;77:24–42. doi: 10.1152/jn.1997.77.1.24. [DOI] [PubMed] [Google Scholar]

[bib61] 61.Ito M., Gilbert C.D. Attention modulates contextual influences in the primary visual cortex of alert monkeys. Neuron. 1999;22:593–604. doi: 10.1016/s0896-6273(00)80713-8. [DOI] [PubMed] [Google Scholar]

[bib62] 62.Maunsell J.H.R. Neuronal representations of cognitive state: reward or attention? Trends Cogn. Sci. 2004;8:261–265. doi: 10.1016/j.tics.2004.04.003. [DOI] [PubMed] [Google Scholar]

[bib63] 63.Krauzlis R.J., Wang L., Yu G., Katz L.N. What is attention? WIREs Cognit. Sci. 2023;14 doi: 10.1002/wcs.1570. [DOI] [PubMed] [Google Scholar]

[bib64] 64.Findling C., Hubert F., Acerbi L., Benson B., Benson J., Birman D., Bonacchi N., Buchanan E.K., Bruijns S., Carandini M., et al. Brain-wide representations of prior information in mouse decision-making. Nature. 2025;645:192–200. doi: 10.1038/s41586-025-09226-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib65] 65.Majumdar A., Ashcroft C., Fritsche M., Tan S., Zatka-Haas P., Folsz O., Walker N., Mistry L., Rominto A.M., Tvrdic M., et al. Distinct representations of economic variables across regions and projections of the frontal cortex. Neuron. 2025;113:4232–4244.e8. doi: 10.1016/j.neuron.2025.09.027. [DOI] [PubMed] [Google Scholar]

[bib66] 66.Fritsche M., Majumdar A., Strickland L., Liebana Garcia S., Bogacz R., Lak A. Temporal regularities shape perceptual decisions and striatal dopamine signals. Nat. Commun. 2024;15:7093. doi: 10.1038/s41467-024-51393-8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib67] 67.Baruni J.K., Lau B., Salzman C.D. Reward expectation differentially modulates attentional behavior and activity in visual area V4. Nat. Neurosci. 2015;18:1656–1663. doi: 10.1038/nn.4141. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib68] 68.Geng J.J., Behrmann M. Spatial probability as an attentional cue in visual search. Percept. Psychophys. 2005;67:1252–1268. doi: 10.3758/bf03193557. [DOI] [PubMed] [Google Scholar]

[bib69] 69.Druker M., Anderson B. Spatial probability AIDS visual stimulus discrimination. Front. Hum. Neurosci. 2010;4 doi: 10.3389/fnhum.2010.00063. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib70] 70.Schnabel U.H., Van der Bijl T., Roelfsema P.R., Lorteije J.A.M. A Direct Comparison of Spatial Attention and Stimulus-Response Compatibility between Mice and Humans. J. Cogn. Neurosci. 2021;33:771–783. doi: 10.1162/jocn_a_01681. [DOI] [PubMed] [Google Scholar]

[bib71] 71.Cohen M.R., Maunsell J.H.R. Attention improves performance primarily by reducing interneuronal correlations. Nat. Neurosci. 2009;12:1594–1600. doi: 10.1038/nn.2439. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib72] 72.Snyder A.C., Yu B.M., Smith M.A. Distinct population codes for attention in the absence and presence of visual stimulation. Nat. Commun. 2018;9:4382. doi: 10.1038/s41467-018-06754-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib73] 73.McAdams C.J., Maunsell J.H.R. Attention to both space and feature modulates neuronal responses in macaque area V4. J. Neurophysiol. 2000;83:1751–1755. doi: 10.1152/jn.2000.83.3.1751. [DOI] [PubMed] [Google Scholar]

[bib74] 74.Connor C.E., Preddie D.C., Gallant J.L., Van Essen D.C. Spatial attention effects in macaque area V4. J. Neurosci. 1997;17:3201–3214. doi: 10.1523/JNEUROSCI.17-09-03201.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib75] 75.McAdams C.J., Reid R.C. Attention modulates the responses of simple cells in monkey primary visual cortex. J. Neurosci. 2005;25:11023–11033. doi: 10.1523/JNEUROSCI.2904-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib76] 76.Steinmetz N.A., Buetfering C., Lecoq J., Lee C.R., Peters A.J., Jacobs E.A.K., Coen P., Ollerenshaw D.R., Valley M.T., de Vries S.E.J., et al. Aberrant Cortical Activity in Multiple GCaMP6-Expressing Transgenic Mouse Lines. eNeuro. 2017;4 doi: 10.1523/ENEURO.0207-17.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib77] 77.Loos M., Koopmans B., Aarts E., Maroteaux G., van der Sluis S., Verhage M., Smit A.B., Smit A.B. Within-strain variation in behavior differs consistently between common inbred strains of mice. Mamm. Genome. 2015;26:348–354. doi: 10.1007/s00335-015-9578-7. [DOI] [PubMed] [Google Scholar]

[bib78] 78.Jaric I., Voelkl B., Amrein I., Wolfer D.P., Novak J., Detotto C., Weber-Stadlbauer U., Meyer U., Manuella F., Mansuy I.M., Würbel H. Using mice from different breeding sites fails to improve replicability of results from single-laboratory studies. Lab Anim (NY) 2024;53:18–22. doi: 10.1038/s41684-023-01307-w. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib79] 79.Voelkl B., Altman N.S., Forsman A., Forstmeier W., Gurevitch J., Jaric I., Karp N.A., Kas M.J., Schielzeth H., Van de Casteele T., Würbel H. Reproducibility of animal research in light of biological variation. Nat. Rev. Neurosci. 2020;21:384–393. doi: 10.1038/s41583-020-0313-3. [DOI] [PubMed] [Google Scholar]

[bib80] 80.Francis N.A., Bohlke K., Kanold P.O. Automated Behavioral Experiments in Mice Reveal Periodic Cycles of Task Engagement within Circadian Rhythms. eNeuro. 2019;6 doi: 10.1523/ENEURO.0121-19.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib81] 81.Tsao C.H., Flint J., Huang G.J. Influence of diurnal phase on behavioral tests of sensorimotor performance, anxiety, learning and memory in mice. Sci. Rep. 2022;12:432. doi: 10.1038/s41598-021-03155-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib82] 82.Georgiou P., Zanos P., Mou T.C.M., An X., Gerhard D.M., Dryanovski D.I., Potter L.E., Highland J.N., Jenne C.E., Stewart B.W., et al. Experimenters' sex modulates mouse behaviors and neural responses to ketamine via corticotropin releasing factor. Nat. Neurosci. 2022;25:1191–1200. doi: 10.1038/s41593-022-01146-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib83] 83.Goldstein S., Wang L., McAlonan K., Torres-Cruz M., Krauzlis R.J. Stimulus-driven visual attention in mice. J. Vis. 2022;22:11. doi: 10.1167/jov.22.1.11. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib84] 84.Funayama K., Hagura N., Ban H., Ikegaya Y. Functional Organization of Flash-Induced V1 Offline Reactivation. J. Neurosci. 2016;36:11727–11738. doi: 10.1523/JNEUROSCI.1575-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib85] 85.Funayama K., Minamisawa G., Matsumoto N., Ban H., Chan A.W., Matsuki N., Murphy T.H., Ikegaya Y. Neocortical Rebound Depolarization Enhances Visual Perception. PLoS Biol. 2015;13 doi: 10.1371/journal.pbio.1002231. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib86] 86.Jin M., Beck J.M., Glickfeld L.L. Neuronal Adaptation Reveals a Suboptimal Decoding of Orientation Tuned Populations in the Mouse Visual Cortex. J. Neurosci. 2019;39:3867–3881. doi: 10.1523/JNEUROSCI.3172-18.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib87] 87.Chandrasekaran A.N., Vermani A., Gupta P., Steinmetz N., Moore T., Sridharan D. Dissociable components of attention exhibit distinct neuronal signatures in primate visual cortex. Sci. Adv. 2024;10 doi: 10.1126/sciadv.adi0645. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib88] 88.Carandini M., Churchland A.K. Probing perceptual decisions in rodents. Nat. Neurosci. 2013;16:824–831. doi: 10.1038/nn.3410. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib89] 89.Steinmetz N.A., Zatka-Haas P., Carandini M., Harris K.D. Distributed coding of choice, action and engagement across the mouse brain. Nature. 2019;576:266–273. doi: 10.1038/s41586-019-1787-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib90] 90.Del Rosario J., Coletta S., Kim S.H., Mobille Z., Peelman K., Williams B., Otsuki A.J., Del Castillo Valerio A., Worden K., Blanpain L.T., et al. Lateral inhibition in V1 controls neural and perceptual contrast sensitivity. Nat. Neurosci. 2025;28:836–847. doi: 10.1038/s41593-025-01888-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib91] 91.Glickfeld L.L., Histed M.H., Maunsell J.H.R. Mouse primary visual cortex is used to detect both orientation and contrast changes. J. Neurosci. 2013;33:19416–19422. doi: 10.1523/JNEUROSCI.3560-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib92] 92.Wolff S.B., Ölveczky B.P. The promise and perils of causal circuit manipulations. Curr. Opin. Neurobiol. 2018;49:84–94. doi: 10.1016/j.conb.2018.01.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib93] 93.Ross L.N., Bassett D.S. Causation in neuroscience: keeping mechanism meaningful. Nat. Rev. Neurosci. 2024;25:81–90. doi: 10.1038/s41583-023-00778-7. [DOI] [PubMed] [Google Scholar]

[bib94] 94.Ruediger S., Scanziani M. Learning speed and detection sensitivity controlled by distinct cortico-fugal neurons in visual cortex. eLife. 2020;9 doi: 10.7554/eLife.59247. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib95] 95.Burgess C.P., Lak A., Steinmetz N.A., Zatka-Haas P., Bai Reddy C., Jacobs E.A.K., Linden J.F., Paton J.J., Ranson A., Schröder S., et al. High-Yield Methods for Accurate Two-Alternative Visual Psychophysics in Head-Fixed Mice. Cell Rep. 2017;20:2513–2524. doi: 10.1016/j.celrep.2017.08.047. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib96] 96.Brainard D.H. The Psychophysics Toolbox. Spat. Vis. 1997;10:433–436. [PubMed] [Google Scholar]

[bib97] 97.Pelli D.G. The VideoToolbox software for visual psychophysics: transforming numbers into movies. Spat. Vis. 1997;10:437–442. [PubMed] [Google Scholar]

[bib98] 98.Bhagat J., Wells M.J., Harris K.D., Carandini M., Burgess C.P. Rigbox: An Open-Source Toolbox for Probing Neurons and Behavior. eNeuro. 2020;7 doi: 10.1523/ENEURO.0406-19.2020. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib99] 99.Williams B., Speed A., Haider B. A novel device for real-time measurement and manipulation of licking behavior in head-fixed mice. J. Neurophysiol. 2018;120:2975–2987. doi: 10.1152/jn.00500.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib100] 100.Mathis A., Mamidanna P., Cury K.M., Abe T., Murthy V.N., Mathis M.W., Bethge M. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat. Neurosci. 2018;21:1281–1289. doi: 10.1038/s41593-018-0209-y. [DOI] [PubMed] [Google Scholar]

[bib101] 101.McGinley M.J., Vinck M., Reimer J., Batista-Brito R., Zagha E., Cadwell C.R., Tolias A.S., Cardin J.A., McCormick D.A. Waking State: Rapid Variations Modulate Neural and Behavioral Responses. Neuron. 2015;87:1143–1161. doi: 10.1016/j.neuron.2015.09.012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib102] 102.Williams B., Del Rosario J., Muzzu T., Peelman K., Coletta S., Bichler E.K., Speed A., Meyer-Baese L., Saleem A.B., Haider B. Spatial modulation of dark versus bright stimulus responses in the mouse visual system. Curr. Biol. 2021;31:4172–4179.e6. doi: 10.1016/j.cub.2021.06.094. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib103] 103.Green D.M., Swets J.A. Wiley; 1966. Signal Detection Theory and Psychophysics. [Google Scholar]

PERMALINK

A standardized and reproducible behavioral framework to elicit multiple aspects of visual spatial attention in mice

Kayla Peelman

Nicole Allen

Jamin Ahn

Joseph Del Rosario

Kaitlin C Jacobson

Esther Kim

Anthony D Lien

Mercedes Lopez-Esteva

Lyndah Lovell

Kendell Worden

Yanlin Yang

James Zhuang

Bilal Haider

Summary

Graphical abstract

Highlights

Motivation

Introduction

Results

Behavioral task design and training phases

Figure 1.

Training protocol with phases enables faster task expertise

Figure 2.

Operant conditioning establishes stimulus-reward association

Stimulus detection across the visual field and at varying contrasts

Figure 3.

Figure 4.

Stimulus location switching elicits spatial attentional effects on behavior

Figure 5.

Standardized training protocol reveals improved contrast sensitivity with attention

Figure 6.

Discussion

Limitations of the study

Resource availability

Lead contact

Materials availability

Data and code availability

Acknowledgments

Author contributions

Declaration of interests

STAR★Methods

Key resources table

Experimental model and study participant details

Method details

Headplate implantation

Handling and habituation

Water restriction and weight monitoring

Behavior apparatus

Eye camera

Task design

Definitions of trials, blocks, sessions, days

Training Phase 0: Operant conditioning

Training Phase 1: Eccentricity shifting

Training Phase 2: Contrast shaping

Training Phase 3: Stimulus location switching

Training Phase 4 (optional): Introduce task-irrelevant bars

Subject inclusion/exclusion

Experimenter observations

Experimenter interventions

Task parameters

Visual stimuli: Gabor gratings

Visual stimuli: Task-irrelevant bar stimuli

Performance metrics and behavioral analysis

Quantification and statistical analysis

Quantification of attention effects

Perceptual contrast sensitivity: psychometric curve fitting

Contrast sensitivity analysis

Psychophysical model

Contrast gain model

Response gain model

c50 and d’ calculation

Contrast response function analysis

Model comparison to dataset

Visualization

Statistical analysis

Footnotes

Supplemental information

References

c₅₀ and d’ calculation