Single-exposure visual memory judgments are reflected in inferotemporal cortex

Contrastive learning explains the emergence and function of visual category-selective regions

Modular and distributed coding theories of category selectivity along the human ventral visual stream have long existed in tension. Here, we present a reconciling framework-contrastive coding-based on a series of analyses relating category selectivity within biological and artificial neural networks. We discover that, in models trained with contrastive self-supervised objectives over a rich natural image diet, category-selective tuning naturally emerges for faces, bodies, scenes, and words. Further, lesions of these model units lead to selective, dissociable recognition deficits, highlighting their distinct functional roles in information processing. Finally, these pre-identified units can predict neural responses in all corresponding face-, scene-, body-, and word-selective regions of human visual cortex, under a highly constrained sparse positive encoding procedure. The success of this single model indicates that brain-like functional specialization can emerge without category-specific learning pressures, as the system learns to untangle rich image content. Contrastive coding, therefore, provides a unifying account of object category emergence and representation in the human brain.

Simple synaptic modulations implement diverse novelty computations

Detecting novelty is ethologically useful for an organism's survival. Recent experiments characterize how different types of novelty over timescales from seconds to weeks are reflected in the activity of excitatory and inhibitory neuron types. Here, we introduce a learning mechanism, familiarity-modulated synapses (FMSs), consisting of multiplicative modulations dependent on presynaptic or pre/postsynaptic neuron activity. With FMSs, network responses that encode novelty emerge under unsupervised continual learning and minimal connectivity constraints. Implementing FMSs within an experimentally constrained model of a visual cortical circuit, we demonstrate the generalizability of FMSs by simultaneously fitting absolute, contextual, and omission novelty effects. Our model also reproduces functional diversity within cell subpopulations, leading to experimentally testable predictions about connectivity and synaptic dynamics that can produce both population-level novelty responses and heterogeneous individual neuron signals. Altogether, our findings demonstrate how simple plasticity mechanisms within a cortical circuit structure can produce qualitatively distinct and complex novelty responses.

Temporal multiplexing of perception and memory codes in IT cortex

A central assumption of neuroscience is that long-term memories are represented by the same brain areas that encode sensory stimuli1. Neurons in inferotemporal (IT) cortex represent the sensory percept of visual objects using a distributed axis code2-4. Whether and how the same IT neural population represents the long-term memory of visual objects remains unclear. Here we examined how familiar faces are encoded in the IT anterior medial face patch (AM), perirhinal face patch (PR) and temporal pole face patch (TP). In AM and PR we observed that the encoding axis for familiar faces is rotated relative to that for unfamiliar faces at long latency; in TP this memory-related rotation was much weaker. Contrary to previous claims, the relative response magnitude to familiar versus unfamiliar faces was not a stable indicator of familiarity in any patch5-11. The mechanism underlying the memory-related axis change is likely intrinsic to IT cortex, because inactivation of PR did not affect axis change dynamics in AM. Overall, our results suggest that memories of familiar faces are represented in AM and perirhinal cortex by a distinct long-latency code, explaining how the same cell population can encode both the percept and memory of faces.

Surprise and novelty in the brain

Notions of surprise and novelty have been used in various experimental and theoretical studies across multiple brain areas and species. However, 'surprise' and 'novelty' refer to different quantities in different studies, which raises concerns about whether these studies indeed relate to the same functionalities and mechanisms in the brain. Here, we address these concerns through a systematic investigation of how different aspects of surprise and novelty relate to different brain functions and physiological signals. We review recent classifications of definitions proposed for surprise and novelty along with links to experimental observations. We show that computational modeling and quantifiable definitions enable novel interpretations of previous findings and form a foundation for future theoretical and experimental studies.

Representations in human primary visual cortex drift over time

Primary sensory regions are believed to instantiate stable neural representations, yet a number of recent rodent studies suggest instead that representations drift over time. To test whether sensory representations are stable in human visual cortex, we analyzed a large longitudinal dataset of fMRI responses to images of natural scenes. We fit the fMRI responses using an image-computable encoding model and tested how well the model generalized across sessions. We found systematic changes in model fits that exhibited cumulative drift over many months. Convergent analyses pinpoint changes in neural responsivity as the source of the drift, while population-level representational dissimilarities between visual stimuli were unchanged. These observations suggest that downstream cortical areas may read-out a stable representation, even as representations within V1 exhibit drift.

Visual recognition in rhesus monkeys requires area TE but not TEO

The primate visual system is often described as a hierarchical feature-conjunction pathway, whereby each level represents an increasingly complex combination of image elements, culminating in the representation of whole coherent images in anterior inferior temporal cortex. Although many models of the ventral visual stream emphasize serial feedforward processing ((Poggio T, Mutch J, Leibo J, Rosasco L, Tacchetti A. The computationalmagic of the ventral stream: sketch of a theory (and why some deep architectures work). TechRep MIT-CSAIL-TR-2012-035. MIT CSAIL, Cambridge, MA. 2012); (Yamins DLK, DiCarlo JJ. Eight open questions in the computational modeling of higher sensory cortex. Curr Opin Neurobiol. 2016:37:114-120.)), anatomical studies show connections that bypass intermediate areas and that feedback to preceding areas ((Distler C, Boussaoud D, Desimone R, Ungerleider LG. Cortical connections of inferior temporal area TEO in macaque monkeys. J Comp Neurol. 1993:334(1):125-150.); (Kravitz DJ, Saleem KS, Baker CI, Mishkin M. A new neural framework for visuospatial processing. Nat Rev Neurosci. 2011:12(4):217-230.)). Prior studies on visual discrimination and object transforms also provide evidence against a strictly feed-forward serial transfer of information between adjacent areas ((Kikuchi R, Iwai E. The locus of the posterior subdivision of the inferotemporal visual learning area in the monkey. Brain Res. 1980:198(2):347-360.); (Weiskrantz L, Saunders RC. Impairments of visual object transforms in monkeys. Brain. 1984:107(4):1033-1072.); (Kar K, DiCarlo JJ. Fast recurrent processing via ventrolateral prefrontal cortex is needed by the primate ventral stream for robust Core visual object recognition. Neuron. 2021:109(1):164-176.e5.)). Thus, we sought to investigate whether behaviorally relevant propagation of visual information is as strictly sequential as sometimes supposed. We compared the accuracy of visual recognition after selective removal of specific subregions of inferior temporal cortex-area TEO, area TE, or both areas combined. Removal of TEO alone had no detectable effect on recognition memory, whereas removal of TE alone produced a large and significant impairment. Combined removal of both areas created no additional deficit relative to removal of TE alone. Thus, area TE is critical for rapid visual object recognition, and detailed image-level visual information can reach area TE via a route other than through TEO.

Multiple traces and altered signal-to-noise in systems consolidation: Evidence from the 7T fMRI Natural Scenes Dataset

How do the neural correlates of recognition change over time? We study natural scene image recognition spanning a year with 7-Tesla functional magnetic resonance imaging (fMRI) of the human brain. We find that the medial temporal lobe (MTL) contribution to recognition persists over 200 d, supporting multiple-trace theory and contradicting a trace transfer (from MTL to cortex) point of view. We then test the hypothesis that the signal-to-noise ratio of traces increases over time, presumably a consequence of synaptic “desaturation” in the weeks following learning. The fMRI trace signature associates with the rate of removal of competing traces and reflects a time-related enhancement of image-feature selectivity. We conclude that multiple MTL traces and improved signal-to-noise may underlie systems-level memory consolidation.

The brain mechanisms of memory consolidation remain elusive. Here, we examine blood-oxygen-level-dependent (BOLD) correlates of image recognition through the scope of multiple influential systems consolidation theories. We utilize the longitudinal Natural Scenes Dataset, a 7-Tesla functional magnetic resonance imaging human study in which ∼135,000 trials of image recognition were conducted over the span of a year among eight subjects. We find that early- and late-stage image recognition associates with both medial temporal lobe (MTL) and visual cortex when evaluating regional activations and a multivariate classifier. Supporting multiple-trace theory (MTT), parts of the MTL activation time course show remarkable fit to a 20-y-old MTT time-dynamical model predicting early trace intensity increases and slight subsequent interference (R² > 0.90). These findings contrast a simplistic, yet common, view that memory traces are transferred from MTL to cortex. Next, we test the hypothesis that the MTL trace signature of memory consolidation should also reflect synaptic “desaturation,” as evidenced by an increased signal-to-noise ratio. We find that the magnitude of relative BOLD enhancement among surviving memories is positively linked to the rate of removal (i.e., forgetting) of competing traces. Moreover, an image-feature and time interaction of MTL and visual cortex functional connectivity suggests that consolidation mechanisms improve the specificity of a distributed trace. These neurobiological effects do not replicate on a shorter timescale (within a session), implicating a prolonged, offline process. While recognition can potentially involve cognitive processes outside of memory retrieval (e.g., re-encoding), our work largely favors MTT and desaturation as perhaps complementary consolidative memory mechanisms.

A neural theory for counting memories

Keeping track of the number of times different stimuli have been experienced is a critical computation for behavior. Here, we propose a theoretical two-layer neural circuit that stores counts of stimulus occurrence frequencies. This circuit implements a data structure, called a count sketch, that is commonly used in computer science to maintain item frequencies in streaming data. Our first model implements a count sketch using Hebbian synapses and outputs stimulus-specific frequencies. Our second model uses anti-Hebbian plasticity and only tracks frequencies within four count categories ("1-2-3-many"), which trades-off the number of categories that need to be distinguished with the potential ethological value of those categories. We show how both models can robustly track stimulus occurrence frequencies, thus expanding the traditional novelty-familiarity memory axis from binary to discrete with more than two possible values. Finally, we show that an implementation of the "1-2-3-many" count sketch exists in the insect mushroom body.

Contribution of Individual Features to Repetition Suppression in Macaque Inferotemporal Cortex

When an image is presented twice in succession, neurons in area TE of macaque inferotemporal cortex exhibit repetition suppression, responding less strongly to the second presentation than to the first. Suppression is known to occur if the adapter and the test image are subtly different from each other. However, it is not known whether cross-suppression occurs between images that are radically different from each other but that share a subset of features. To explore this issue, we measured repetition suppression using colored shapes. On interleaved trials, the test image might be identical to the adapter, might share its shape or color alone or might differ from it totally. At the level of the neuronal population as a whole, suppression was especially deep when adapter and test were identical, intermediate when they shared only one attribute and minimal when they shared neither attribute. At the level of the individual neuron, the degree of suppression depended not only on the properties of the two images but also on the preferences of the neuron. Suppression was deeper when the repeated color or shape was preferred by the neuron than when it was not. This effect might arise from feature-specific adaptation or alternatively from adapter-induced fatigue. Both mechanisms conform to the principle that the degree of suppression is determined by the preferences of the neuron.

Meta-learning synaptic plasticity and memory addressing for continual familiarity detection

Over the course of a lifetime, we process a continual stream of information. Extracted from this stream, memories must be efficiently encoded and stored in an addressable manner for retrieval. To explore potential mechanisms, we consider a familiarity detection task in which a subject reports whether an image has been previously encountered. We design a feedforward network endowed with synaptic plasticity and an addressing matrix, meta-learned to optimize familiarity detection over long intervals. We find that anti-Hebbian plasticity leads to better performance than Hebbian plasticity and replicates experimental results such as repetition suppression. A combinatorial addressing function emerges, selecting a unique neuron as an index into the synaptic memory matrix for storage or retrieval. Unlike previous models, this network operates continuously and generalizes to intervals it has not been trained on. Our work suggests a biologically plausible mechanism for continual learning and demonstrates an effective application of machine learning for neuroscience discovery.

Novel stimuli evoke excess activity in the mouse primary visual cortex

Rapid detection and processing of stimulus novelty are key elements of adaptive behavior. Predictive coding theories postulate that novel stimuli should be encoded differently from familiar stimuli. Here, we show that the majority of neurons in layer 2/3 of the mouse primary visual cortex exhibit a significant excess response to novel visual stimuli. The distinction between novel and familiar images developed rapidly, requiring only a few repeated presentations. We show that this phenomenon can be described by a model of cascading adaptation. This ubiquitous mechanism makes it likely that similar computations could be carried out in many brain areas.

To explore how neural circuits represent novel versus familiar inputs, we presented mice with repeated sets of images with novel images sparsely substituted. Using two-photon calcium imaging to record from layer 2/3 neurons in the mouse primary visual cortex, we found that novel images evoked excess activity in the majority of neurons. This novelty response rapidly emerged, arising with a time constant of 2.6 ± 0.9 s. When a new image set was repeatedly presented, a majority of neurons had similarly elevated activity for the first few presentations, which decayed to steady state with a time constant of 1.4 ± 0.4 s. When we increased the number of images in the set, the novelty response’s amplitude decreased, defining a capacity to store ∼15 familiar images under our conditions. These results could be explained quantitatively using an adaptive subunit model in which presynaptic neurons have individual tuning and gain control. This result shows that local neural circuits can create different representations for novel versus familiar inputs using generic, widely available mechanisms.

A self-supervised domain-general learning framework for human ventral stream representation

Anterior regions of the ventral visual stream encode substantial information about object categories. Are top-down category-level forces critical for arriving at this representation, or can this representation be formed purely through domain-general learning of natural image structure? Here we present a fully self-supervised model which learns to represent individual images, rather than categories, such that views of the same image are embedded nearby in a low-dimensional feature space, distinctly from other recently encountered views. We find that category information implicitly emerges in the local similarity structure of this feature space. Further, these models learn hierarchical features which capture the structure of brain responses across the human ventral visual stream, on par with category-supervised models. These results provide computational support for a domain-general framework guiding the formation of visual representation, where the proximate goal is not explicitly about category information, but is instead to learn unique, compressed descriptions of the visual world.

Completing the puzzle: Why studies in non-human primates are needed to better understand the effects of non-invasive brain stimulation

Brain stimulation is a core method in neuroscience. Numerous non-invasive brain stimulation (NIBS) techniques are currently in use in basic and clinical research, and recent advances promise the ability to non-invasively access deep brain structures. While encouraging, there is a surprising gap in our understanding of precisely how NIBS perturbs neural activity throughout an interconnected network, and how such perturbed neural activity ultimately links to behaviour. In this review, we will consider why non-human primate (NHP) models of NIBS are ideally situated to address this gap in knowledge, and why the oculomotor network that moves our line of sight offers a particularly valuable platform in which to empirically test hypothesis regarding NIBS-induced changes in brain and behaviour. NHP models of NIBS will enable investigation of the complex, dynamic effects of brain stimulation across multiple hierarchically interconnected brain areas, networks, and effectors. By establishing such links between brain and behavioural output, work in NHPs can help optimize experimental and therapeutic approaches, improve NIBS efficacy, and reduce side-effects of NIBS.

Remembering the Past to See the Future

In addition to the role that our visual system plays in determining what we are seeing right now, visual computations contribute in important ways to predicting what we will see next. While the role of memory in creating future predictions is often overlooked, efficient predictive computation requires the use of information about the past to estimate future events. In this article, we introduce a framework for understanding the relationship between memory and visual prediction and review the two classes of mechanisms that the visual system relies on to create future predictions. We also discuss the principles that define the mapping from predictive computations to predictive mechanisms and how downstream brain areas interpret the predictive signals computed by the visual system. Expected final online publication date for the Annual Review of Vision Science, Volume 7 is September 2021. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.

Neural evidence for recognition of naturalistic videos in monkey hippocampus

The role of the hippocampus in recognition memory has long been a source of debate. Tasks used to study recognition that typically require an explicit probe, where the participant must make a response to prove they remember, yield mixed results on hippocampal involvement. Here, we tasked monkeys to freely view naturalistic videos, and only tested their memory via looking times for two separate novel versus repeat video conditions on each trial. Notably, a large proportion (>30%) of hippocampal neurons differentiated these videos via changes in firing rates time-locked to the duration of their presentation on screen, and not during the delay period between them as would be expected for working memory. Many of these single neurons (>15%) contributed to both retrieval conditions, and differentiated novel from repeat videos across trials with trial-unique content, suggesting they detect familiarity. The majority of neurons contributing to the classifier showed an enhancement in firing rate on repeat compared with novel videos, a pattern which has not previously been shown in hippocampus. These results suggest the hippocampus contributes to recognition memory via familiarity during free-viewing.

Pinpointing the neural signatures of single-exposure visual recognition memory

Memories of the images that we have seen are thought to be reflected in the reduction of neural responses in high-level visual areas such as inferotemporal (IT) cortex, a phenomenon known as repetition suppression (RS). We challenged this hypothesis with a task that required rhesus monkeys to report whether images were novel or repeated while ignoring variations in contrast, a stimulus attribute that is also known to modulate the overall IT response. The monkeys' behavior was largely contrast invariant, contrary to the predictions of an RS-inspired decoder, which could not distinguish responses to images that are repeated from those that are of lower contrast. However, the monkeys' behavioral patterns were well predicted by a linearly decodable variant in which the total spike count was corrected for contrast modulation. These results suggest that the IT neural activity pattern that best aligns with single-exposure visual recognition memory behavior is not RS but rather sensory referenced suppression: reductions in IT population response magnitude, corrected for sensory modulation.

Evaluating the evidence for expectation suppression in the visual system

Reports of expectation suppression have shaped the development of influential predictive coding-based theories of visual perception. However recent work has highlighted confounding factors that may mimic or inflate expectation suppression effects. In this review, we describe four confounds that are prevalent across experiments that tested for expectation suppression: effects of surprise, attention, stimulus repetition and adaptation, and stimulus novelty. With these confounds in mind we then critically review the evidence for expectation suppression across probabilistic cueing, statistical learning, oddball, action-outcome learning and apparent motion designs. We found evidence for expectation suppression within a specific subset of statistical learning designs that involved weeks of sequence learning prior to neural activity measurement. Across other experimental contexts, whereby stimulus appearance probabilities were learned within one or two testing sessions, there was inconsistent evidence for genuine expectation suppression. We discuss how an absence of expectation suppression could inform models of predictive processing, repetition suppression and perceptual decision-making. We also provide suggestions for designing experiments that may better test for expectation suppression in future work.

A temporal record of the past with a spectrum of time constants in the monkey entorhinal cortex

Episodic memory is believed to be intimately related to our experience of the passage of time. Indeed, neurons in the hippocampus and other brain regions critical to episodic memory code for the passage of time at a range of timescales. The origin of this temporal signal, however, remains unclear. Here, we examined temporal responses in the entorhinal cortex of macaque monkeys as they viewed complex images. Many neurons in the entorhinal cortex were responsive to image onset, showing large deviations from baseline firing shortly after image onset but relaxing back to baseline at different rates. This range of relaxation rates allowed for the time since image onset to be decoded on the scale of seconds. Further, these neurons carried information about image content, suggesting that neurons in the entorhinal cortex carry information about not only when an event took place but also, the identity of that event. Taken together, these findings suggest that the primate entorhinal cortex uses a spectrum of time constants to construct a temporal record of the past in support of episodic memory.

Magnitude, time course, and specificity of rapid adaptation across mouse visual areas

Adaptation is a ubiquitous feature of sensory processing whereby recent experience shapes future responses. The mouse primary visual cortex (V1) is particularly sensitive to recent experience, where a brief stimulus can suppress subsequent responses for seconds. This rapid adaptation profoundly impacts perception, suggesting that its effects are propagated along the visual hierarchy. To understand how rapid adaptation influences sensory processing, we measured its effects at key nodes in the visual system: in V1, three higher visual areas (HVAs: lateromedial, anterolateral, and posteromedial), and the superior colliculus (SC) in awake mice of both sexes using single-unit recordings. Consistent with the feed-forward propagation of adaptation along the visual hierarchy, we find that neurons in layer 4 adapt less strongly than those in other layers of V1. Furthermore, neurons in the HVAs adapt more strongly, and recover more slowly, than those in V1. The magnitude and time course of adaptation was comparable in each of the HVAs and in the SC, suggesting that adaptation may not linearly accumulate along the feed-forward visual processing hierarchy. Despite the increase in adaptation in the HVAs compared with V1, the effects were similarly orientation specific across all areas. These data reveal that adaptation profoundly shapes cortical processing, with increasing impact at higher levels in the cortical hierarchy, and also strongly influencing computations in the SC. Thus, we find robust, brain-wide effects of rapid adaptation on sensory processing.NEW & NOTEWORTHY Rapid adaptation dynamically alters sensory signals to account for recent experience. To understand how adaptation affects sensory processing and perception, we must determine how it impacts the diverse set of cortical and subcortical areas along the hierarchy of the mouse visual system. We find that rapid adaptation strongly impacts neurons in primary visual cortex, the higher visual areas, and the colliculus, consistent with its profound effects on behavior.

Visual novelty, curiosity, and intrinsic reward in machine learning and the brain

A strong preference for novelty emerges in infancy and is prevalent across the animal kingdom. When incorporated into reinforcement-based machine learning algorithms, visual novelty can act as an intrinsic reward signal that vastly increases the efficiency of exploration and expedites learning, particularly in situations where external rewards are difficult to obtain. Here we review parallels between recent developments in novelty-driven machine learning algorithms and our understanding of how visual novelty is computed and signaled in the primate brain. We propose that in the visual system, novelty representations are not configured with the principal goal of detecting novel objects, but rather with the broader goal of flexibly generalizing novelty information across different states in the service of driving novelty-based learning.

Population response magnitude variation in inferotemporal cortex predicts image memorability

Most accounts of image and object encoding in inferotemporal cortex (IT) focus on the distinct patterns of spikes that different images evoke across the IT population. By analyzing data collected from IT as monkeys performed a visual memory task, we demonstrate that variation in a complementary coding scheme, the magnitude of the population response, can largely account for how well images will be remembered. To investigate the origin of IT image memorability modulation, we probed convolutional neural network models trained to categorize objects. We found that, like the brain, different natural images evoked different magnitude responses from these networks, and in higher layers, larger magnitude responses were correlated with the images that humans and monkeys find most memorable. Together, these results suggest that variation in IT population response magnitude is a natural consequence of the optimizations required for visual processing, and that this variation has consequences for visual memory.

Mechanisms underlying sharpening of visual response dynamics with familiarity

Experience-dependent modifications of synaptic connections are thought to change patterns of network activities and stimulus tuning with learning. However, only a few studies explored how synaptic plasticity shapes the response dynamics of cortical circuits. Here, we investigated the mechanism underlying sharpening of both stimulus selectivity and response dynamics with familiarity observed in monkey inferotemporal cortex. Broadening the distribution of activities and stronger oscillations in the response dynamics after learning provide evidence for synaptic plasticity in recurrent connections modifying the strength of positive feedback. Its interplay with slow negative feedback via firing rate adaptation is critical in sharpening response dynamics. Analysis of changes in temporal patterns also enables us to disentangle recurrent and feedforward synaptic plasticity and provides a measure for the strengths of recurrent synaptic plasticity. Overall, this work highlights the importance of analyzing changes in dynamics as well as network patterns to further reveal the mechanisms of visual learning.

Human frontoparietal cortex represents behaviorally relevant target status based on abstract object features

Searching for items that are useful given current goals, or "target" recognition, requires observers to flexibly attend to certain object properties at the expense of others. This could involve focusing on the identity of an object while ignoring identity-preserving transformations such as changes in viewpoint or focusing on its current viewpoint while ignoring its identity. To effectively filter out variation due to the irrelevant dimension, performing either type of task is likely to require high-level, abstract search templates. Past work has found target recognition signals in areas of ventral visual cortex and in subregions of parietal and frontal cortex. However, target status in these tasks is typically associated with the identity of an object, rather than identity-orthogonal properties such as object viewpoint. In this study, we used a task that required subjects to identify novel object stimuli as targets according to either identity or viewpoint, each of which was not predictable from low-level properties such as shape. We performed functional MRI in human subjects of both sexes and measured the strength of target-match signals in areas of visual, parietal, and frontal cortex. Our multivariate analyses suggest that the multiple-demand (MD) network, including subregions of parietal and frontal cortex, encodes information about an object's status as a target in the relevant dimension only, across changes in the irrelevant dimension. Furthermore, there was more target-related information in MD regions on correct compared with incorrect trials, suggesting a strong link between MD target signals and behavior. NEW & NOTEWORTHY Real-world target detection tasks, such as searching for a car in a crowded parking lot, require both flexibility and abstraction. We investigated the neural basis of these abilities using a task that required invariant representations of either object identity or viewpoint. Multivariate decoding analyses of our whole brain functional MRI data reveal that invariant target representations are most pronounced in frontal and parietal regions, and the strength of these representations is associated with behavioral performance.

Current and future goals are represented in opposite patterns in object-selective cortex

Adaptive behavior requires the separation of current from future goals in working memory. We used fMRI of object-selective cortex to determine the representational (dis)similarities of memory representations serving current and prospective perceptual tasks. Participants remembered an object drawn from three possible categories as the target for one of two consecutive visual search tasks. A cue indicated whether the target object should be looked for first (currently relevant), second (prospectively relevant), or if it could be forgotten (irrelevant). Prior to the first search, representations of current, prospective and irrelevant objects were similar, with strongest decoding for current representations compared to prospective (Experiment 1) and irrelevant (Experiment 2). Remarkably, during the first search, prospective representations could also be decoded, but revealed anti-correlated voxel patterns compared to currently relevant representations of the same category. We propose that the brain separates current from prospective memories within the same neuronal ensembles through opposite representational patterns.