Adaptive behavior in complex environments critically relies on the ability to appropriately link specific choices or actions to their outcomes. However, the neural mechanisms that support the ability to credit only those past choices believed to have caused the observed outcomes remain unclear. Here, we leverage multivariate pattern analyses of functional magnetic resonance imaging (fMRI) data and an adaptive learning task to shed light on the underlying neural mechanisms of such specific credit assignment. We find that the lateral orbitofrontal cortex (lOFC) and hippocampus (HC) code for the causal choice identity when credit needs to be assigned for choices that are separated from outcomes by a long delay, even when this delayed transition is punctuated by interim decisions. Further, we show when interim decisions must be made, learning is additionally supported by lateral frontopolar cortex (lFPC). Our results indicate that lFPC holds previous causal choices in a "pending" state until a relevant outcome is observed, and the fidelity of these representations predicts the fidelity of subsequent causal choice representations in lOFC and HC during credit assignment. Together, these results highlight the importance of the timely reinstatement of specific causes in lOFC and HC in learning choice-outcome relationships when delays and choices intervene, a critical component of real-world learning and decision making.

Many objects in the real world have features that vary over time, creating uncertainty in how they will look in the future. This uncertainty makes statistical knowledge about the likelihood of features critical to attention demanding processes such as visual search. However, little is known about how the uncertainty of visual features is integrated into predictions about search targets in the brain. In the current study, we test the idea that regions prefrontal cortex code statistical knowledge about search targets before the onset of search. Across 20 human participants (13 female; 7 male), we observe target identity in the multivariate pattern and uncertainty in the overall activation of dorsolateral prefrontal cortex (DLPFC) and inferior frontal junction (IFJ) in advance of the search display. This indicates that the target identity (mean) and uncertainty (variance) of the target distribution are coded independently within the same regions. Furthermore, once the search display appears the univariate IFJ signal scaled with the distance of the actual target from the expected mean, but more so when expected variability was low. These results inform neural theories of attention by showing how the prefrontal cortex represents both the identity and expected variability of features in service of top-down attentional control.SIGNIFICANCE STATEMENT Theories of attention and working memory posit that when we engage in complex cognitive tasks our performance is determined by how precisely we remember task-relevant information. However, in the real world the properties of objects change over time, creating uncertainty about many aspects of the task. There is currently a gap in our understanding of how neural systems represent this uncertainty and combine it with target identity information in anticipation of attention demanding cognitive tasks. In this study, we show that the prefrontal cortex represents identity and uncertainty as unique codes before task onset. These results advance theories of attention by showing that the prefrontal cortex codes both target identity and uncertainty to implement top-down attentional control.

Visual attention is often characterized as being guided by precise memories for target objects. However, real-world search targets have dynamic features that vary over time, meaning that observers must predict how the target could look based on how features are expected to change. Despite its importance, little is known about how target feature predictions influence feature-based attention, or how these predictions are represented in the target template. In Experiment 1 (N = 60 university students), we show observers readily track the statistics of target features over time and adapt attentional priority to predictions about the distribution of target features. In Experiments 2a and 2b (N = 480 university students), we show that these predictions are encoded into the target template as a distribution of likelihoods over possible target features, which are independent of memory precision for the cued item. These results provide a novel demonstration of how observers represent predicted feature distributions when target features are uncertain and show that these predictions are used to set attentional priority during visual search. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Theories of attention often focus on how known, definitive visual features are represented prior to the onset of a task. However, objects in the real-world are dynamic and have features that change overtime, creating uncertainty in their future appearance. This means that observers must make predictions about how the target could look. Representing the uncertainty of target features is essential for predictive processes, yet how the brain represents the uncertainty of upcoming visual stimuli is unclear. In this study, we investigate the brain regions that encode the distribution of possible target features prior to the onset of a visual search task. In each trial, participants (N=7) heard a cue-tone (high- or low-pitch) which corresponded to the central tendency of a target color (pink-orange, or blue-green, respectively). However, the true target color was drawn from a Gaussian distribution centered on the predicted color with either high or low variability, depending on the block. In each trial, there was one distractor object with a randomly selected color. Thus, to optimally represent the target object participants had to know both the central tendency of the target color, as well as the distribution of possible colors the target could have (i.e., the uncertainty). Using a combination of linear decoders and representational similarity analysis, we show the regions of early visual cortex and rostro-lateral prefrontal cortex (RLPFC) carry representations of the predicted target and the uncertainty of those predictions. Further, univariate analyses at the time of search found that RLPFC, medial frontal gyrus, and intraparietal cortex also coded prediction errors for the target object which depended on the variability of the target distribution. These results suggest that a network of sensory and frontal regions carry joint codes for both the mean and uncertainty of the distribution of possible upcoming search targets.

Animals abstract compact representations of a task's structure, which supports accelerated learning and flexible behavior. Whether and how such abstracted representations may be used to assign credit for inferred, but unobserved, relationships in structured environments are unknown. We develop a hierarchical reversal-learning task and Bayesian learning model to assess the computational and neural mechanisms underlying how humans infer specific choice-outcome associations via structured knowledge. We find that the medial prefrontal cortex (mPFC) efficiently represents hierarchically related choice-outcome associations governed by the same latent cause, using a generalized code to assign credit for both experienced and inferred outcomes. Furthermore, the mPFC and lateral orbitofrontal cortex track the current "position" within a latent association space that generalizes over stimuli. Collectively, these findings demonstrate the importance of both tracking the current position in an abstracted task space and efficient, generalizable representations in the prefrontal cortex for supporting flexible learning and inference in structured environments.

Recent research on attentional templates suggests that representations of target items held in WM are not static representations of past stimuli, but dynamic representations that prioritize task-relevant information in the environment. However, complex environments may obfuscate which source of information is best to prioritize, meaning the visual system must make judgments about the predictability of information (information value) to efficiently allocate attention. How these these predictions are generated and how they relate to the allocation of attention are still poorly understood. To determine the relationship between information value and feature based attention, we designed an online search task for a target object defined by an orientation and color. At the start of each trial, participants (N=240) were cued with the most likely features of the target, but knew these features could change to any within a distribution of possible values. Target color was always sampled from a distribution with high uncertainty (SD=55), but the target orientation was drawn either low-variability (SD=10), medium-variability (SD=25), or high-variability (SD=40) distribution. A separate group of participants were allocated to a control condition each feature was drawn from identical t-distributions with low-variability (SD=10). Interleaved were 18 probe trials which asked participants to rate the likelihood of possible targets, gauging participants knowledge of the underlying feature distribution. Results showed that attention to each target target orientation was enhanced when its relative information value was high, while color was suppressed, mirroring participants knowledge of each feature distribution. Attention to either feature returned to baseline when the information values were approximately equal. These results point to a critical role of information value in feature based prioritization in the attentional template.

Attention operates as a cognitive gate that selects sensory information for entry into memory and awareness (Driver, 2001, British Journal of Psychology, 92, 53–78). Under many circumstances, the selected information is task-relevant and important to remember, but sometimes perceptually salient nontarget objects will capture attention and enter into awareness despite their irrelevance (Adams & Gaspelin, 2020, Attention, Perception, & Psychophysics, 82[4], 1586–1598). Recent studies have shown that repeated exposures with salient distractor will diminish their ability to capture attention, but the relationship between suppression and later cognitive processes such as memory and awareness remains unclear. If learned attentional suppression (indicated by reduced capture costs) occurs at the sensory level and prevents readout to other cognitive processes, one would expect memory and awareness to dimmish commensurate with improved suppression. Here, we test this hypothesis by measuring memory precision and awareness of salient nontargets over repeated exposures as capture costs decreased. Our results show that stronger learned suppression is accompanied by reductions in memory precision and confidence in having seen a color singleton at all, suggesting that such suppression operates at the sensory level to prevent further processing of the distractor object.

The orbital frontal cortex (OFC) has long been linked to goal-directed, flexible behaviors. Recent evidence suggests the OFC plays key roles in representing the abstracted structure of task spaces, and using this representation for flexible inferences during both learning and choice. Here, we review convergent evidence from studies in animal models and humans in support of this view. We begin by considering early accounts of OFC function, then discuss how more recent evidence supports theories that have re-cast OFC's function as representing the structure of a task or environment for flexible inference. Finally, we turn to neural recording studies that provide insights into the underlying representations and computations the OFC may implement in coordination with other brain areas. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Recent research on working-memory (WM) suggests that representations held in WM are not necessarily static representations of past data, but can serve as dynamic representations of expected information. Nobre and Stokes (2019) recently introduced the term “premembering” to describe this proactive role for memory in attention. If true, then it should be possible to dissociate the remembered representation of a previously seen “cue” stimulus from the “premembered” expectation of a future target by statistically manipulating what will happen during search. We asked subjects to search for a target RDK after seeing a “cue” with a specific color and direction of motion (Supplemental S1). The target varied from the cue in predictable ways: one feature dimension (e.g., motion) was drawn from a distribution narrowly centered over the cued feature (low-variance dimension) and the other feature (e.g., color) was sampled from a broad distribution (high-variance dimension). The standard deviation of the low-variance distribution changed over the experiment. In Experiment 1, there were additional “probe” trials on which subjects reported the remembered features of the cue using a color or motion wheel. Analysis of RTs showed that subjects were sensitive to changes in the variance of the low-variance dimension but, probe responses showed WM representations were not sensitive to these changes (S2). Experiment 2 used the same design but “probe” trials asked subjects to predict the target features. This tested whether expectations factor into predictive representations rather than changing the content of memory items. We replicated the RT results from Experiment 1 showing that response-times were sensitive changes in variance, but additionally show that the precision of probe responses changed in step with the variance of the low-variance distribution (S3). These results show a fundamental distinction between how expectations about upcoming sensory data factor into premembered versus remembered representations.

Real-world visual search targets are frequently imperfect perceptual matches to our internal templates. For example, a friend on different occasions will have different clothes, hairstyles, and accessories, but some of these may vary more than others. The ability to deal with template-to-target variability is important to visual search in natural environments, but we know relatively little about how this is handled by the attentional system. Here, we test the hypothesis that top-down attentional biases are sensitive to the variance of target features and prioritize less-variable dimensions. Subjects were shown target cues composed of coloured dots moving in a specific direction followed by a working memory probe (30%) or visual search display (70%). Critically, the target features in the visual search display differed from the cue, with one feature drawn from a narrow distribution (low-variance dimension), and the other sampled from a broader distribution (high-variance dimension). The results demonstrate that subjects used knowledge of the likely cue-to-target variance to set template precision and bias attentional selection. Our results suggest that observers are sensitive to the variance of feature dimensions within a target and use this information to weight mechanisms of attentional selection.

All models of attention include the concept of an attentional template (or a target or search template). The template is conceptualized as target information held in memory that is used for prioritizing sensory processing and determining if an object matches the target. It is frequently assumed that the template contains a veridical copy of the target. However, we review recent evidence showing that the template encodes a version of the target that is adapted to the current context (e.g. distractors, task, etc.); information held within the template may include only a subset of target features, real world knowledge, pre-existing perceptual biases, or even be a distorted version of the veridical target. We argue that the template contents are customized in order to maximize the ability to prioritize information that distinguishes targets from distractors. We refer to this as template-to-distractor distinctiveness and hypothesize that it contributes to visual search efficiency by exaggerating target-to-distractor dissimilarity.

Introduction: Research shows the visual system efficiently encodes peripheral objects as statistical representations, known as ensembles. However, few studies have explored the role of ensembles in visual search. Some models suggest that attention is drawn to ensembles with average qualities similar to the target (Im et al. 2015). Other models propose that learned ensemble-target associations facilitate visual search by cueing the location of the target. (Alvarez, 2011). Our project examines the function of ensembles to understand how they are used to facilitate target search and localization. Methods: Participants (N=20 per experiment) located a target line in one of two groups of lines, which formed ensembles in opposite locations on the screen. The average orientation of one ensemble matched the target orientation. The non-matching ensemble was 30 to 60 degrees different. Participants reported whether the target was in the left- or right-side ensemble, or was absent. In Experiment 1, the target was equally likely to be in all locations. In Experiment 2, the target was in the matching ensemble on 75% of trials and in either the non-matching ensemble or absent in 25% of trials. Results: Results from both experiments indicated that participants made significantly more initial saccades toward the matching ensemble, suggesting that the ensembles captured attention. Only in Experiment 2 did participants have faster response times, suggesting that participants used ensembles as cues to the target location after learning the ensemble-target association. This is further supported by evidence that participants were less likely to check the opposite ensemble after finding the target. This pattern suggests ensembles primarily influence visual search by acting as learned cues to the targets location. Conclusion: These results suggest that target-matching ensembles capture attention, but the effect on visual search is small unless there is a meaningful association between the ensemble and the target. Meeting abstract presented at VSS 2017