Biology subjects

Gershman, S. J.

Publications and source records attributed to Gershman, S. J..

13 recordsLinked to original sources

bioRxiv2018

Finding structure in multi-armed bandits

How do humans search for rewards? This question is commonly studied using multi-armed bandit tasks, which require participants to trade off exploration and exploitation. Standard multi-armed bandits assume that each option has an independent reward distribution. However, learning about options independently is unrealistic, since in the real world options often share an underlying structure. We study a class of structured bandit tasks, which we use to probe how generalization guides exploration. In a structured multi-armed bandit, options have a correlation structure dictated by a latent function. We focus on bandits in which rewards are linear functions of an options spatial position. Across 5 experiments, we find evidence that participants utilize functional structure to guide their exploration, and also exhibit a learning-to-learn effect across rounds, becoming progressively faster at identifying the latent function. Our experiments rule out several heuristic explanations and show that the same findings obtain with non-linear functions. Comparing several models of learning and decision making, we find that the best model of human behavior in our tasks combines three computational mechanisms: (1) function learning, (2) clustering of reward distributions across rounds, and (3) uncertainty-guided exploration. Our results suggest that human reinforcement learning can utilize latent structure in sophisticated ways to improve efficiency.

animal behavior and cognition↗

bioRxiv2018

Learning to act by integrating mental simulations and physical experiments

People can learn about the effects of their actions either by performing physical experiments or by running mental simulations. Physical experiments are reliable but risky; mental simulations are unreliable but safe. We investigate how people negotiate the balance between these strategies. Participants attempted to shoot a ball at a target, and could pay to take practice shots (physical experiments). They could also simply think (run mental simulations), but were incentivized to act quickly by paying for time. We demonstrate that the amount of thinking time and physical experiments is sensitive to trial characteristics in a way that is consistent with a model that integrates information across simulation and experimentation and decides online when to perform each.

animal behavior and cognition↗

bioRxiv2018

Uncertainty and exploration

In order to discover the most rewarding actions, agents must collect information about their environment, potentially foregoing reward. The optimal solution to this \"explore-exploit\" dilemma is often computationally challenging, but principled algorithmic approximations exist. These approximations utilize uncertainty about action values in different ways. Some random exploration algorithms scale the level of choice stochasticity with the level of uncertainty. Other directed exploration algorithms add a \"bonus\" to action values with high uncertainty. Random exploration algorithms are sensitive to total uncertainty across actions, whereas directed exploration algorithms are sensitive to relative uncertainty. This paper reports a multi-armed bandit experiment in which total and relative uncertainty were orthogonally manipulated. We found that humans employ both exploration strategies, and that these strategies are independently controlled by different uncertainty computations.

animal behavior and cognition↗

bioRxiv2017

Rethinking dopamine prediction errors

Midbrain dopamine neurons are commonly thought to report a reward prediction error, as hypothesized by reinforcement learning theory. While this theory has been highly successful, several lines of evidence suggest that dopamine activity also encodes sensory prediction errors unrelated to reward. Here we develop a new theory of dopamine function that embraces a broader conceptualization of prediction errors. By signaling errors in both sensory and reward predictions, dopamine supports a form of reinforcement learning that lies between model-based and model-free algorithms. This account remains consistent with current canon regarding the correspondence between dopamine transients and reward prediction errors, while also accounting for new data suggesting a role for these signals in phenomena such as sensory preconditioning and identity unblocking, which ostensibly draw upon knowledge beyond reward predictions.

neuroscience↗

bioRxiv2017

Remembrance of Inferences Past

Bayesian models of cognition assume that people compute probability distributions over hypotheses. However, the required computations are frequently intractable or prohibitively expensive. Since people often encounter many closely related distributions, selective reuse of computations (amortized inference) is a computationally efficient use of the brains limited resources. We present three experiments that provide evidence for amortization in human probabilistic reasoning. When sequentially answering two related queries about natural scenes, participants responses to the second query systematically depend on the structure of the first query. This influence is sensitive to the content of the queries, only appearing when the queries are related. Using a cognitive load manipulation, we find evidence that people amortize summary statistics of previous inferences, rather than storing the entire distribution. These findings support the view that the brain trades off accuracy and computational cost, to make efficient use of its limited cognitive resources to approximate probabilistic inference.

animal behavior and cognition↗

bioRxiv2017

Decision by sampling implements efficient coding of psychoeconomic functions

The theory of decision by sampling (DbS) proposes that an attributes subjective value is its rank within a sample of attribute values retrieved from memory. This can account for instances of context dependence beyond the reach of classic theories which assume stable preferences. In this paper, we provide a normative justification for DbS that is based on the principle of efficient coding. The efficient representation of information in a noiseless communication channel is characterized by a uniform response distribution, which the rank transformation implements. However, cognitive limitations imply that decision samples are finite, introducing noise. Efficient coding in a noisy channel requires smoothing of the signal, a principle that leads to a new generalization of DbS. This generalization is closely connected to range-frequency theory, and helps descriptively account for a wider set of behavioral observations, such as how context sensitivity varies with the number of available response categories.

animal behavior and cognition↗

bioRxiv2017

Dopamine, Inference, and Uncertainty

The hypothesis that the phasic dopamine response reports a reward prediction error has become deeply entrenched. However, dopamine neurons exhibit several notable deviations from this hypothesis. A coherent explanation for these deviations can be obtained by analyzing the dopamine response in terms of Bayesian reinforcement learning. The key idea is that prediction errors are modulated by probabilistic beliefs about the relationship between cues and outcomes, updated through Bayesian inference. This account can explain dopamine responses to inferred value in sensory preconditioning, the effects of cue pre-exposure (latent inhibition) and adaptive coding of prediction errors when rewards vary across orders of magnitude. We further postulate that orbitofrontal cortex transforms the stimulus representation through recurrent dynamics, such that a simple error-driven learning rule operating on the transformed representation can implement the Bayesian reinforcement learning update.

neuroscience↗

bioRxiv2017

Using computational theory to constrain statistical models of neural data

Computational neuroscience is, to first order, dominated by two approaches: the \"bottom-up\" approach, which searches for statistical patterns in large-scale neural recordings, and the \"top-down\" approach, which begins with a theory of computation and considers plausible neural implementations. While this division is not clear-cut, we argue that these approaches should be much more intimately linked. From a Bayesian perspective, computational theories provide constrained prior distributions on neural data--albeit highly sophisticated ones. By connecting theory to observation via a probabilistic model, we provide the link necessary to test, evaluate, and revise our theories in a data-driven and statistically rigorous fashion. This review highlights examples of this theory-driven pipeline for neural data analysis in recent literature and illustrates it with a worked example based on the temporal difference learning model of dopamine.

neuroscience↗

bioRxiv2017

Pure Correlates of Exploration and Exploitation in the Human Brain

Balancing exploration and exploitation is a fundamental problem in reinforcement learning. Previous neuroimaging studies of the exploration-exploitation dilemma could not completely disentangle these two processes, making it difficult to unambiguously identify their neural signatures. We overcome this problem using a task in which subjects can either observe (pure exploration) or bet (pure exploitation). Insula and dorsal anterior cingulate cortex showed significantly greater activity on observe trials compared to bet trials, suggesting that these regions play a role in driving exploration. A model-based analysis of task performance suggested that subjects chose to observe until a critical evidence threshold was reached. We observed a neural signature of this evidence accumulation process in ventromedial prefrontal cortex. These findings support theories positing an important role for anterior cingulate cortex in exploration, while also providing a new perspective on the roles of insula and ventromedial prefrontal cortex.\n\nSignificance StatementSitting down at a familiar restaurant, you may choose to order an old favorite or sample a new dish. In reinforcement learning theory, this is known as the exploration-exploitation dilemma. The optimal solution is known to be intractable; therefore, humans must use heuristic strategies. Behavioral studies have revealed several candidate strategies, but identifying the neural mechanisms underlying these strategies is complicated due to the fact that exploration and exploitation are not perfectly dissociable in standard tasks. Using an \"observe or bet\" task, we identify for the first time pure neural correlates of exploration and exploitation in the human brain.

neuroscience↗

bioRxiv2016

The hippocampus as a predictive map

A cognitive map has long been the dominant metaphor for hippocampal function, embracing the idea that place cells encode a geometric representation of space. However, evidence for predictive coding, reward sensitivity, and policy dependence in place cells suggests that the representation is not purely spatial. We approach this puzzle from a reinforcement learning perspective: what kind of spatial representation is most useful for maximizing future reward? We show that the answer takes the form of a predictive representation. This representation captures many aspects of place cell responses that fall outside the traditional view of a cognitive map. Furthermore, we argue that entorhinal grid cells encode a low-dimensional basis set for the predictive representation, useful for suppressing noise in predictions and extracting multiscale structure for hierarchical planning.

neuroscience↗

bioRxiv2016

Compositional Inductive Biases in Function Learning

How do people recognize and learn about complex functional structure? Taking inspiration from other areas of cognitive science, we propose that this is achieved by harnessing compositionality: complex structure is decomposed into simpler building blocks. We formalize this idea within the framework of Bayesian regression using a grammar over Gaussian process kernels, and compare this approach with other structure learning approaches. Participants consistently chose compositional (over non-compositional) extrapolations and interpolations of functions. Experiments designed to elicit priors over functional patterns revealed an inductive bias for compositional structure. Compositional functions were perceived as subjectively more predictable than non-compositional functions, and exhibited other signatures of predictability, such as enhanced memorability and reduced numerosity. Taken together, these results support the view that the human intuitive theory of functions is inherently compositional.

animal behavior and cognition↗

bioRxiv2016

The successor representation in human reinforcement learning

Theories of reward learning in neuroscience have focused on two families of algorithms, thought to capture deliberative vs. habitual choice. \"Model-based\" algorithms compute the value of candidate actions from scratch, whereas \"model-free\" algorithms make choice more efficient but less flexible by storing pre-computed action values. We examine an intermediate algorithmic family, the successor representation (SR), which balances flexibility and efficiency by storing partially computed action values: predictions about future events. These pre-computation strategies differ in how they update their choices following changes in a task. SRs reliance on stored predictions about future states predicts a unique signature of insensitivity to changes in the tasks sequence of events, but flexible adjustment following changes to rewards. We provide evidence for such differential sensitivity in two behavioral studies with humans. These results suggest that the SR is a computational substrate for semi-flexible choice in humans, introducing a subtler, more cognitive notion of habit.

neuroscience↗

bioRxiv2016

Predictive representations can link model-based reinforcement learning to model-free mechanisms

Humans and animals are capable of evaluating actions by considering their long-run future rewards through a process described using model-based reinforcement learning (RL) algorithms. The mechanisms by which neural circuits perform the computations prescribed by model-based RL remain largely unknown; however, multiple lines of evidence suggest that neural circuits supporting model-based behavior are structurally homologous to and overlapping with those thought to carry out model-free temporal difference (TD) learning. Here, we lay out a family of approaches by which model-based computation may be built upon a core of TD learning. The foundation of this framework is the successor representation, a predictive state representation that, when combined with TD learning of value predictions, can produce a subset of the behaviors associated with model-based learning, while requiring less decision-time computation than dynamic programming. Using simulations, we delineate the precise behavioral capabilities enabled by evaluating actions using this approach, and compare them to those demonstrated by biological organisms. We then introduce two new algorithms that build upon the successor representation while progressively mitigating its limitations. Because this framework can account for the full range of observed putatively model-based behaviors while still utilizing a core TD framework, we suggest that it represents a neurally plausible family of mechanisms for model-based evaluation.\n\nAuthor SummaryAccording to standard models, when confronted with a choice, animals and humans rely on two separate, distinct processes to come to a decision. One process deliberatively evaluates the consequences of each candidate action and is thought to underlie the ability to flexibly come up with novel plans. The other process gradually increases the propensity to perform behaviors that were previously successful and is thought to underlie automatically executed, habitual reflexes. Although computational principles and animal behavior support this dichotomy, at the neural level, there is little evidence supporting a clean segregation. For instance, although dopamine -- famously implicated in drug addiction and Parkinsons disease -- currently only has a well-defined role in the automatic process, evidence suggests that it also plays a role in the deliberative process. In this work, we present a computational framework for resolving this mismatch. We show that the types of behaviors associated with either process could result from a common learning mechanism applied to different strategies for how populations of neurons could represent candidate actions. In addition to demonstrating that this account can produce the full range of flexible behavior observed in the empirical literature, we suggest experiments that could detect the various approaches within this framework.

neuroscience↗