Biology subjects

Fusi, S.

Publications and source records attributed to Fusi, S..

At least 37 records · Page 2Linked to original sources

bioRxiv2023

Perirhinal Cortex Learns A Predictive Map of The Task Environment

Goal-directed tasks involve acquiring an internal model, known as a predictive map, of relevant stimuli and associated outcomes to guide behavior. Here, we identified neural signatures of a predictive map of task behavior in perirhinal cortex (Prh). Mice learned to perform a tactile working memory task by classifying sequential whisker stimuli over multiple training stages. Chemogenetic inactivation demonstrated that Prh is involved in task learning. Chronic two-photon calcium imaging, population analysis, and computational modeling revealed that Prh encodes stimulus features as sensory prediction errors. Prh forms stable stimulus-outcome associations that expand in a retrospective manner and generalize as animals learn new contingencies. Stimulus-outcome associations are linked to prospective network activity encoding possible expected outcomes. This link is mediated by cholinergic signaling to guide task performance, demonstrated by acetylcholine imaging and perturbation. We propose that Prh combines error-driven and map-like properties to acquire a predictive map of learned task behavior.

neuroscience↗

bioRxiv2022

Neural representational geometry correlates with behavioral differences between monkeys

Animals likely use a variety of strategies to solve laboratory tasks. Traditionally, combined analysis of behavioral and neural recording data across subjects employing different strategies may obscure important signals and give confusing results. Hence, it is essential to develop techniques that can infer strategy at the single-subject level. We analyzed an experiment in which two monkeys performed a visually cued rule-based task. The analysis of their performance shows no indication that they used a different strategy. However, when we examined the geometry of stimulus representations in the state space of the neural activities recorded in dorsolateral prefrontal cortex, we found striking differences between the two monkeys. Our purely neural results induced us to reanalyze the behavior. The new analysis showed that the differences in representational geometry correlate with differences in the reaction times, revealing behavioral differences we were unaware of. All these analyses indicate that the monkeys are using different strategies. Finally, using recurrent neural network models trained to perform the same task, we show that these strategies correlate with the amount of training, suggesting a possible explanation for the observed neural and behavioral differences.

neuroscience↗

bioRxiv2022

The geometry of hippocampal CA2 representations enables abstract coding of social familiarity and identity

Social recognition consists of multiple memory processes, including the detection of familiarity - the ability to rapidly distinguish familiar from novel individuals - and recollection - the effortful recall of where a social episode occurred and who was present. At present, the neural mechanisms for these different social memory processes remain unknown. Here, we investigate the population code for novel and familiar individuals in mice using calcium imaging of neural activity in a region crucial for social memory, the dorsal CA2 area of the hippocampus. We report that familiarity changes CA2 representations of social encounters to meet the different demands of social memory. While novel individuals are represented in a low-dimensional geometry that allows for rapid generalization, familiar individuals are represented in a higher-dimensional geometry that supports high-capacity memory storage. The magnitude of the change in dimensionality of CA2 representations for a given individual predicts the performance of that individual in a social recognition memory test, suggesting a direct relationship between the representational geometry and memory-guided behavior. Finally, we show that familiarity is encoded as an abstract variable with neural responses generalizing across different identities and spatial locations. Thus, through the tuning of the geometry of structured neural activity, CA2 is able to meet the complex demands of multiple social memory processes.

neuroscience↗

bioRxiv2021

Abstract representations emerge naturally in neural networks trained to perform multiple tasks

Humans and other animals demonstrate a remarkable ability to generalize knowledge across distinct contexts and objects during natural behavior. We posit that this ability to generalize arises from a specific representational geometry, that we call abstract and that is referred to as disentangled in machine learning. These abstract representations have been observed in recent neurophysiological studies. However, it is unknown how they emerge. Here, using feedforward neural networks, we demonstrate that the learning of multiple tasks causes abstract representations to emerge, using both supervised and reinforcement learning. We show that these abstract representations enable few-sample learning and reliable generalization on novel tasks. We conclude that abstract representations of sensory and cognitive variables may emerge from the multiple behaviors that animals exhibit in the natural world, and, as a consequence, could be pervasive in high-level brain regions. We also make several specific predictions about which variables will be represented abstractly.

neuroscience↗

bioRxiv2021

Signatures of rapid synaptic learning in the hippocampus during novel experiences

Neurons in the hippocampus exhibit striking selectivity for specific combinations of sensory features, forming representations which are thought to subserve episodic memory. Even during a completely novel experience, ensembles of hippocampal "place cells" are rapidly configured such that the population sparsely encodes visited locations, stabilizing within minutes of the first exposure to a new environment. What cellular mechanisms enable this fast encoding of experience? Here we leverage virtual reality and large scale neural recordings to dissect the effects of novelty and experience on the dynamics of place field formation. We show that the place fields of many CA1 neurons transiently shift locations and modulate the amplitude of their activity immediately after place field formation, consistent with rapid plasticity mechanisms driven by plateau potentials and somatic burst spiking. These motifs were particularly enriched during initial exploration of a novel context and decayed with experience. Our data suggest that novelty modulates the effective learning rate in CA1, favoring burst-driven field formation to support fast synaptic updating during new experience.

neuroscience↗

bioRxiv2021

The geometry of face memory

The ability to recognize familiar visual objects is critical to survival. A central assumption of neuroscience is that long-term memories are represented by the same brain areas that encode sensory stimuli (1). Neurons in inferotemporal (IT) cortex represent the sensory percept of visual objects using a distributed axis code (2-4). 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 face patch AM and perirhinal cortex. We found that familiar faces were represented in a distinct subspace from unfamiliar faces. The familiar face subspace was shifted relative to the unfamiliar face subspace at short latency and then distorted to increase neural distances between familiar faces at long latency. This distortion enabled markedly improved discrimination of familiar faces in both AM and PR. Inactivation of PR did not affect these memory traces in AM, suggesting that the memory traces arise from intrinsic recurrent processes within IT cortex or interactions with downstream regions outside the medial temporal lobe (5, 6). Overall, our results reveal that memories of familiar faces are represented in IT and perirhinal cortex by a distinct long-latency code that is optimized to distinguish familiar identities.

neuroscience↗

bioRxiv2021

The non-linear mixed representations in somatosensory cortex support simple and complex tasks

Neural responses are often highly heterogeneous non-linear functions of multiple task variables, a signature of a high-dimensional geometry of the neural representations. We studied the representational geometry in the somatosensory cortex of mice trained to report the curvature of objects using their whiskers. High-speed videos of the whisker movements revealed that the task can be solved by linearly integrating multiple whisker contacts over time. However, the neural activity in somatosensory cortex reflects a process of non-linear integration of spatio-temporal features of the sensory inputs. Although the responses at first appear disorganized, we could identify an interesting structure in the representational geometry: different whisker contacts are disentangled variables represented in approximately, but not fully, orthogonal subspaces of the neural activity space. The observed geometry allows linear readouts to perform a broad class of tasks of different complexities without compromising the ability to generalize to novel situations.

neuroscience↗

bioRxiv2020

The sensorimotor strategies and neuronal representations of tactile shape discrimination in mice

Humans and other animals can identify objects by active touch, requiring the coordination of exploratory motion and tactile sensation. The brain integrates movements with the resulting tactile signals to form a holistic representation of object identity. We developed a shape discrimination task that challenged head-fixed mice to discriminate concave from convex shapes. Behavioral decoding revealed that mice did this by comparing contacts across whiskers. In contrast, mice performing a shape detection task simply summed up contacts over whiskers. We recorded populations of neurons in the barrel cortex, which processes whisker input, to identify how it encoded the corresponding sensorimotor variables. Neurons across the cortical layers encoded touch, whisker motion, and task-related signals. Sensory representations were task-specific: during shape discrimination, neurons responded most robustly to behaviorally relevant whiskers, overriding somatotopy. We suggest a similar dynamic modulation may underlie object recognition in other brain areas and species.

neuroscience↗

bioRxiv2019

Face familiarity detection with complex synapses

Synaptic plasticity is a complex phenomenon involving multiple biochemical processes that operate on different timescales. We recently showed that this complexity can greatly increase the memory capacity of neural networks when the variables that characterize the synaptic dynamics have limited precision, as in biological systems. These types of complex synapses have been tested mostly on simple memory retrieval problems involving random and uncorrelated patterns. Here we turn to a real-world problem, face familiarity detection, and we show that also in this case it is possible to take advantage of synaptic complexity to store in memory a large number of faces that can be recognized at a later time. In particular, we show that the familiarity memory capacity of a system with complex synapses grows almost linearly with the number of the synapses and quadratically with the number of neurons. Complex synapses are superior to simple ones, which are characterized by a single variable, even when the total number of dynamical variables is matched. We further show that complex and simple synapses have distinct signatures that are testable in proposed experiments. Our results indicate that a memory system with complex synapses can be used in real-world tasks such as face familiarity detection. SignificanceThe complexity of biological synapses is probably important for enabling us to remember the past for a long time and rapidly store new memories. The advantage of complex synapses in terms of memory capacity is significant when the variables that characterize the synaptic dynamics have limited precision. This advantage has been estimated under the simplifying assumption that the memories to be stored are random and uncorrelated. Here we show that synaptic complexity is important also in a more challenging and realistic face familiarity detection task. We built a simple neural circuit that can report whether a face has been previously seen or not. This circuit incorporates complex synapses that operate on multiple timescales. The memory performance of this circuit is significantly higher than in the case in which synapses are simple, indicating that the complexity of biological synapses can be important also in real-world memory tasks.

neuroscience↗

bioRxiv2019

Coding of social novelty in the hippocampal CA2 region and its disruption and rescue in a mouse model of schizophrenia

The hippocampal CA2 region is essential for social memory and has been implicated in neuropsychiatric disorders. However, little is known about how CA2 neural activity encodes social interactions and how this coding is altered in disease. We recorded from CA2 pyramidal neurons as mice engaged in social interactions and found that while CA2 failed to stably represent spatial location, CA2 activity encoded contextual changes and novel social stimuli. In the Df(16)A+/- mouse model of the human 22q11.2 microdeletion, a major schizophrenia risk factor, CA2 activity showed a surprising increase in spatial coding while failing to encode social novelty, consistent with the social memory deficit in these mice. Previous work has shown that CA2 pyramidal neurons are hyperpolarized in Df(16)A+/- mice, likely as a result of upregulation of TREK-1 K+ current. We found that administration of a TREK-1 antagonist rescued the social memory deficits and restored normal CA2 coding properties in Df(16)A+/- mice, supporting a crucial role for CA2 in the encoding of novel social stimuli and social dysfunction.

neuroscience↗

bioRxiv2019

Flexible recruitment of memory-based choice representations by human medial-frontal cortex

Decisions in complex environments rely on flexibly utilizing past experience as required by context and instructions1. This process depends on the medial frontal cortex (MFC) and the medial temporal lobe (MTL)2-5, but it remains unknown how these structures jointly implement flexible memory retrieval6,7. We recorded single neurons in MFC and MTL while human subjects switched8 between making memory- and categorization-based decisions9,10. Here we show that MFC rapidly implements changing task demands by utilizing different subspaces of neural activity during different types of decisions. In contrast, no effect of task demands was seen in the MTL. Choices requiring memory retrieval selectively engaged phase-locking of MFC neurons to field potentials in the theta-frequency band in the MTL. Choice-selective neurons in MFC signaled abstract yes-no decisions independent of behavioral response modality (button press or saccade). These findings reveal a novel mechanism for flexibly and selectively engaging memory retrieval11-14 and show that unlike perceptual decision-making15, memory-related information is only represented in frontal cortex when choices require it.

neuroscience↗

bioRxiv2019

Are place cells just memory cells? Memory compression leads to spatial tuning and history dependence

The observation of place cells has suggested that the hippocampus plays a special role in encoding spatial information. However, place cell responses are modulated by several non-spatial variables, and reported to be rather unstable. Here we propose a memory model of the hippocampus that provides a novel interpretation of place cells consistent with these observations. We hypothesize that the hippocampus is a memory device that takes advantage of the correlations between sensory experiences to generate compressed representations of the episodes that are stored in memory. A simple neural network model that can efficiently compress information naturally produces place cells that are similar to those observed in experiments. It predicts that the activity of these cells is variable and that the fluctuations of the place fields encode information about the recent history of sensory experiences. Place cells may simply be a consequence of a memory compression process implemented in the hippocampus. Significance StatementNumerous studies on humans revealed the importance of the hippocampus in memory formation. The rodent literature instead focused on the spatial representations that are observed in navigation experiments. Here we propose a simple model of the hippocampus that reconciles the main findings of the human and rodent studies. The model assumes that the hippocampus is a memory system that generates compressed representations of sensory experiences using previously acquired knowledge about the statistics of the world. These experiences can then be memorized more efficiently. The sensory experiences during the exploration of an environment, when compressed by the hippocampus, lead naturally to spatial representations similar to those observed in rodent studies and to the emergence of place cells.

neuroscience↗

bioRxiv2019

Hippocampal network reorganization underlies the formation of a temporal association memory

Episodic memory requires linking events in time, a function dependent on the hippocampus. In \"trace\" fear conditioning, animals learn to associate a neutral cue with an aversive stimulus despite their separation in time by a delay period on the order of tens of seconds. But how this temporal association forms remains unclear. Here we use 2-photon calcium imaging to track neural population dynamics over the complete time-course of learning and show that, in contrast to previous theories, the hippocampus does not generate persistent activity to bridge the time delay. Instead, learning is concomitant with broad changes in the active neural population in CA1. While neural responses were highly stochastic in time, cue identity could be reliably read out from population activity rates over longer timescales after learning. These results question the ubiquity of neural sequences during temporal association learning, and suggest that trace fear conditioning relies on mechanisms that differ from persistent activity accounts of working memory.

neuroscience↗

bioRxiv2018

Delay activity dynamics: task dependent time encoding and low dimensional trajectories

Our decisions often depend on multiple sensory experiences separated by time delays. The brain can remember these experiences and, simultaneously, estimate the timing between events. To understand the mechanisms underlying working memory and time encoding we analyze neural activity recorded during delays in four experiments on non-human primates. To disambiguate potential mechanisms, we propose two analyses, namely, decoding the passage of time from neural data, and computing the cumulative dimensionality of the neural trajectory over time. Time can be decoded with high precision in tasks where timing information is relevant and with lower precision when irrelevant for performing the task. Neural trajectories are always observed to be low dimensional. These constraints rule out working memory models that rely on constant, sustained activity, and neural networks with high dimensional trajectories, like reservoir networks. Instead, recurrent networks trained with backpropagation capture the time encoding properties and the dimensionality observed in the data.

neuroscience↗

bioRxiv2018

The geometry of abstraction in hippocampus and prefrontal cortex

The curse of dimensionality plagues models of reinforcement learning and decision-making. The process of abstraction solves this by constructing abstract variables describing features shared by different specific instances, reducing dimensionality and enabling generalization in novel situations. Here we characterized neural representations in monkeys performing a task where a hidden variable described the temporal statistics of stimulus-response-outcome mappings. Abstraction was defined operationally using the generalization performance of neural decoders across task conditions not used for training. This type of generalization requires a particular geometric format of neural representations. Neural ensembles in dorsolateral pre-frontal cortex, anterior cingulate cortex and hippocampus, and in simulated neural networks, simultaneously represented multiple hidden and explicit variables in a format reflecting abstraction. Task events engaging cognitive operations modulated this format. These findings elucidate how the brain and artificial systems represent abstract variables, variables critical for generalization that in turn confers cognitive flexibility.

neuroscience↗

bioRxiv2018

Flexible coding of memory and space in the primate hippocampus during virtual navigation

Hippocampal maps of space change across tasks. The mechanisms of this effect remain unclear. To examine this, we recorded activity of hippocampal neurons in monkeys navigating the same virtual maze during two different tasks: a foraging task requiring only cue guided navigation, and a memory task also requiring context-object association. Within each task, individual neurons had spatially-selective response fields, enabling a linear classifier to decode position in the virtual environment in each task. However, the population code did not generalize across tasks. This was due to sensory and mnemonic coding of non-spatial features and their associations by single neurons during each period of the associative memory task. Thus, sensory and mnemonic representations of non-spatial features shape maps of space in the primate hippocampus during virtual navigation. This may reflect a fundamental role of the hippocampus in compressing information from a variety of sources for efficient memory storage.

neuroscience↗

bioRxiv2018

A distributed neural code in ensembles of dentate gyrus granule cells

The tuning properties of neurons in a given brain region have been traditionally viewed as the under-pinnings of computation in neural circuits. However, at the higher levels of processing, specialization is often elusive, instead a mix of sensory, cognitive and behavioural quantities drive neural activity. In such networks, ensembles of neurons, rather than single units with easily interpretable tuning properties, encode behaviourally relevant variables. Here we show that this is the case also in the dentate gyrus and CA1 subregions of the hippocampus. Using calcium imaging in freely moving mice, we decoded the instantaneous position, direction of motion and speed from the activity of hundreds of cells in the hippocampus of mice freely exploring an arena. For the vast majority of neurons in both regions, their response properties were not predictive of their importance for encoding position. Furthermore, we could decode position from populations of cells that were important for decoding direction of motion and vice versa, showing that these quantities are encoded by largely overlapping ensembles as in distributed neural code. Finally, we found that correlated activities had an impact on decoding performance in CA1 but not in dentate gyrus, suggesting different enconding strategies for these areas. Our analysis indicates that classical methods of analysis based on single cell response properties might be insufficient to accurately characterize the neural computation in a given area. In contrast, population analysis may help highlight previously overlooked properties of hippocampal circuits.

neuroscience↗

bioRxiv2017

Classifiers with limited connectivity

For many neural network models in which neurons are trained to classify inputs like perceptrons, the number of inputs that can be classified is limited by the connectivity of each neuron, even when the total number of neurons is very large. This poses the problem of how the biological brain can take advantage of its huge number of neurons given that the connectivity is sparse. One solution is to combine multiple perceptrons together, as in committee machines. The number of classifiable random patterns would then grow linearly with the number of perceptrons, even when each perceptron has limited connectivity. However, the problem is moved to the downstream readout neurons, which would need a number of connections that is as large as the number of perceptrons. Here we propose a different approach in which the readout is implemented by connecting multiple perceptrons in a recurrent attractor neural network. We prove analytically that the number of classifiable random patterns can grow unboundedly with the number of perceptrons, even when the connectivity of each perceptron remains finite. Most importantly, both the recurrent connectivity and the connectivity of downstream readouts also remain finite. Our study shows that feed-forward neural classifiers with numerous long range afferent connections can be replaced by recurrent networks with sparse long range connectivity without sacrificing the classification performance. Our strategy could be used to design more general scalable network architectures with limited connectivity, which resemble more closely the brain neural circuits which are dominated by recurrent connectivity.

neuroscience↗