Generative replay underlies compositional inference in the hippocampal-prefrontal circuit.

Human reasoning depends on reusing pieces of information by putting them together in new ways. However, very little is known about how compositional computation is implemented in the brain. Here, we ask participants to solve a series of problems that each require constructing a whole from a set of elements. With fMRI, we find that representations of novel constructed objects in the frontal cortex and hippocampus are relational and compositional. With MEG, we find that replay assembles elements into compounds, with each replay sequence constituting a hypothesis about a possible configuration of elements. The content of sequences evolves as participants solve each puzzle, progressing from predictable to uncertain elements and gradually converging on the correct configuration. Together, these results suggest a computational bridge between apparently distinct functions of hippocampal-prefrontal circuitry and a role for generative replay in compositional inference and hypothesis testing.

Impaired neural replay of inferred relationships in schizophrenia.

An ability to build structured mental maps of the world underpins our capacity to imagine relationships between objects that extend beyond experience. In rodents, such representations are supported by sequential place cell reactivations during rest, known as replay. Schizophrenia is proposed to reflect a compromise in structured mental representations, with animal models reporting abnormalities in hippocampal replay and associated ripple activity during rest. Here, utilizing magnetoencephalography (MEG), we tasked patients with schizophrenia and control participants to infer unobserved relationships between objects by reorganizing visual experiences containing these objects. During a post-task rest session, controls exhibited fast spontaneous neural reactivation of presented objects that replayed inferred relationships. Replay was coincident with increased ripple power in hippocampus. Patients showed both reduced replay and augmented ripple power relative to controls, convergent with findings in animal models. These abnormalities are linked to impairments in behavioral acquisition and subsequent neural representation of task structure.

Human Replay Spontaneously Reorganizes Experience.

Knowledge abstracted from previous experiences can be transferred to aid new learning. Here, we asked whether such abstract knowledge immediately guides the replay of new experiences. We first trained participants on a rule defining an ordering of objects and then presented a novel set of objects in a scrambled order. Across two studies, we observed that representations of these novel objects were reactivated during a subsequent rest. As in rodents, human "replay" events occurred in sequences accelerated in time, compared to actual experience, and reversed their direction after a reward. Notably, replay did not simply recapitulate visual experience, but followed instead a sequence implied by learned abstract knowledge. Furthermore, each replay contained more than sensory representations of the relevant objects. A sensory code of object representations was preceded 50 ms by a code factorized into sequence position and sequence identity. We argue that this factorized representation facilitates the generalization of a previously learned structure to new objects.

A distributional code for value in dopamine-based reinforcement learning.

Since its introduction, the reward prediction error theory of dopamine has explained a wealth of empirical phenomena, providing a unifying framework for understanding the representation of reward and value in the brain1-3. According to the now canonical theory, reward predictions are represented as a single scalar quantity, which supports learning about the expectation, or mean, of stochastic outcomes. Here we propose an account of dopamine-based reinforcement learning inspired by recent artificial intelligence research on distributional reinforcement learning4-6. We hypothesized that the brain represents possible future rewards not as a single mean, but instead as a probability distribution, effectively representing multiple future outcomes simultaneously and in parallel. This idea implies a set of empirical predictions, which we tested using single-unit recordings from mouse ventral tegmental area. Our findings provide strong evidence for a neural realization of distributional reinforcement learning.

Investigating associations of delay discounting with brain structure, working memory, and episodic memory.

INTRODUCTION: Delay discounting (DD), the preference for smaller and sooner rewards over larger and later ones, is an important behavioural phenomenon for daily functioning of increasing interest within psychopathology. The neurobiological mechanisms behind DD are not well understood and the literature on structural correlates of DD shows inconsistencies. METHODS: Here we leveraged a large openly available dataset (n = 1196) to investigate associations with memory performance and gray and white matter correlates of DD using linked independent component analysis. RESULTS: Greater DD was related to smaller anterior temporal gray matter volume. Associations of DD with total cortical volume, subcortical volumes, markers of white matter microscopic organization, working memory, and episodic memory scores were not significant after controlling for education and income. CONCLUSION: Effects of size comparable to the one we identified would be unlikely to be replicated with sample sizes common in many previous studies in this domain, which may explain the incongruities in the literature. The paucity and small size of the effects detected in our data underscore the importance of using large samples together with methods that accommodate their statistical structure and appropriate control for confounders, as well as the need to devise paradigms with improved task parameter reliability in studies relating brain structure and cognitive abilities with DD.

Dynamic diversity is the answer to proxy failure.

We argue that a diverse and dynamic pool of agents mitigates proxy failure. Proxy modularity plays a key role in the ongoing production of diversity. We review examples from a range of scales.

Agent-specific learning signals for self-other distinction during mentalising.

Humans have a remarkable ability to simulate the minds of others. How the brain distinguishes between mental states attributed to self and mental states attributed to someone else is unknown. Here, we investigated how fundamental neural learning signals are selectively attributed to different agents. Specifically, we asked whether learning signals are encoded in agent-specific neural patterns or whether a self-other distinction depends on encoding agent identity separately from this learning signal. To examine this, we tasked subjects to learn continuously 2 models of the same environment, such that one was selectively attributed to self and the other was selectively attributed to another agent. Combining computational modelling with magnetoencephalography (MEG) enabled us to track neural representations of prediction errors (PEs) and beliefs attributed to self, and of simulated PEs and beliefs attributed to another agent. We found that the representational pattern of a PE reliably predicts the identity of the agent to whom the signal is attributed, consistent with a neural self-other distinction implemented via agent-specific learning signals. Strikingly, subjects exhibiting a weaker neural self-other distinction also had a reduced behavioural capacity for self-other distinction and displayed more marked subclinical psychopathological traits. The neural self-other distinction was also modulated by social context, evidenced in a significantly reduced decoding of agent identity in a nonsocial control task. Thus, we show that self-other distinction is realised through an encoding of agent identity intrinsic to fundamental learning signals. The observation that the fidelity of this encoding predicts psychopathological traits is of interest as a potential neurocomputational psychiatric biomarker.

Computing Value from Quality and Quantity in Human Decision-Making.

How organisms learn the value of single stimuli through experience is well described. In many decisions, however, value estimates are computed "on the fly" by combining multiple stimulus attributes. The neural basis of this computation is poorly understood. Here we explore a common scenario in which decision-makers must combine information about quality and quantity to determine the best option. Using fMRI, we examined the neural representation of quality, quantity, and their integration into an integrated subjective value signal in humans of both genders. We found that activity within inferior frontal gyrus (IFG) correlated with offer quality, while activity in the intraparietal sulcus (IPS) specifically correlated with offer quantity. Several brain regions, including the anterior cingulate cortex (ACC), were sensitive to an interaction of quality and quantity. However, the ACC was uniquely activated by quality, quantity, and their interaction, suggesting that this region provides a substrate for flexible computation of value from both quality and quantity. Furthermore, ACC signals across subjects correlated with the strength of quality and quantity signals in IFG and IPS, respectively. ACC tracking of subjective value also correlated with choice predictability. Finally, activity in the ACC was elevated for choice trials, suggesting that ACC provides a nexus for the computation of subjective value in multiattribute decision-making.SIGNIFICANCE STATEMENT Would you prefer three apples or two oranges? Many choices we make each day require us to weigh up the quality and quantity of different outcomes. Using fMRI, we show that option quality is selectively represented in the inferior frontal gyrus, while option quantity correlates with areas of the intraparietal sulcus that have previously been associated with numerical processing. We show that information about the two is integrated into a value signal in the anterior cingulate cortex, and the fidelity of this integration predicts choice predictability. Our results demonstrate how on-the-fly value estimates are computed from multiple attributes in human value-based decision-making.

Parsing the Role of the Hippocampus in Approach-Avoidance Conflict.

The hippocampus plays a central role in the approach-avoidance conflict that is central to the genesis of anxiety. However, its exact functional contribution has yet to be identified. We designed a novel gambling task that generated approach-avoidance conflict while controlling for spatial processing. We fit subjects' behavior using a model that quantified the subjective values of choice options, and recorded neural signals using functional magnetic resonance imaging (fMRI). Distinct functional signals were observed in anterior hippocampus, with inferior hippocampus selectively recruited when subjects rejected a gamble, to a degree that covaried with individual differences in anxiety. The superior anterior hippocampus, in contrast, uniquely demonstrated value signals that were potentiated in the context of approach-avoidance conflict. These results implicate the anterior hippocampus in behavioral avoidance and choice monitoring, in a manner relevant to understanding its role in anxiety. Our findings highlight interactions between subregions of the hippocampus as an important focus for future study.

Local striatal reward signals can be predicted from corticostriatal connectivity.

A defining feature of the basal ganglia is their anatomical organization into multiple cortico-striatal loops. A central tenet of this architecture is the idea that local striatal function is determined by its precise connectivity with cortex, creating a functional topography that is mirrored within cortex and striatum. Here we formally test this idea using both human anatomical and functional imaging, specifically asking whether within striatal subregions one can predict between-voxel differences in functional signals based on between-voxel differences in corticostriatal connectivity. We show that corticostriatal connectivity profiles predict local variation in reward signals in bilateral caudate nucleus and putamen, expected value signals in bilateral caudate nucleus, and response effector activity in bilateral putamen. These data reveal that, even within individual striatal regions, local variability in corticostriatal anatomical connectivity predicts functional differentiation.

Harm to others outweighs harm to self in moral decision making.

Concern for the suffering of others is central to moral decision making. How humans evaluate others' suffering, relative to their own suffering, is unknown. We investigated this question by inviting subjects to trade off profits for themselves against pain experienced either by themselves or an anonymous other person. Subjects made choices between different amounts of money and different numbers of painful electric shocks. We independently varied the recipient of the shocks (self vs. other) and whether the choice involved paying to decrease pain or profiting by increasing pain. We built computational models to quantify the relative values subjects ascribed to pain for themselves and others in this setting. In two studies we show that most people valued others' pain more than their own pain. This was evident in a willingness to pay more to reduce others' pain than their own and a requirement for more compensation to increase others' pain relative to their own. This "hyperaltruistic" valuation of others' pain was linked to slower responding when making decisions that affected others, consistent with an engagement of deliberative processes in moral decision making. Subclinical psychopathic traits correlated negatively with aversion to pain for both self and others, in line with reports of aversive processing deficits in psychopathy. Our results provide evidence for a circumstance in which people care more for others than themselves. Determining the precise boundaries of this surprisingly prosocial disposition has implications for understanding human moral decision making and its disturbance in antisocial behavior.

Model-Based Reasoning in Humans Becomes Automatic with Training.

Model-based and model-free reinforcement learning (RL) have been suggested as algorithmic realizations of goal-directed and habitual action strategies. Model-based RL is more flexible than model-free but requires sophisticated calculations using a learnt model of the world. This has led model-based RL to be identified with slow, deliberative processing, and model-free RL with fast, automatic processing. In support of this distinction, it has recently been shown that model-based reasoning is impaired by placing subjects under cognitive load--a hallmark of non-automaticity. Here, using the same task, we show that cognitive load does not impair model-based reasoning if subjects receive prior training on the task. This finding is replicated across two studies and a variety of analysis methods. Thus, task familiarity permits use of model-based reasoning in parallel with other cognitive demands. The ability to deploy model-based reasoning in an automatic, parallelizable fashion has widespread theoretical implications, particularly for the learning and execution of complex behaviors. It also suggests a range of important failure modes in psychiatric disorders.

Anterior cingulate cortex instigates adaptive switches in choice by integrating immediate and delayed components of value in ventromedial prefrontal cortex.

Actions can lead to an immediate reward or punishment and a complex set of delayed outcomes. Adaptive choice necessitates the brain track and integrate both of these potential consequences. Here, we designed a sequential task whereby the decision to exploit or forego an available offer was contingent on comparing immediate value and a state-dependent future cost of expending a limited resource. Crucially, the dynamics of the task demanded frequent switches in policy based on an online computation of changing delayed consequences. We found that human subjects choose on the basis of a near-optimal integration of immediate reward and delayed consequences, with the latter computed in a prefrontal network. Within this network, anterior cingulate cortex (ACC) was dynamically coupled to ventromedial prefrontal cortex (vmPFC) when adaptive switches in choice were required. Our results suggest a choice architecture whereby interactions between ACC and vmPFC underpin an integration of immediate and delayed components of value to support flexible policy switching that accommodates the potential delayed consequences of an action.

Dopamine Regulates Approach-Avoidance in Human Sensation-Seeking.

BACKGROUND: Sensation-seeking is a trait that constitutes an important vulnerability factor for a variety of psychopathologies with high social cost. However, little is understood either about the mechanisms underlying motivation for intense sensory experiences or their neuropharmacological modulation in humans. METHODS: Here, we first evaluate a novel paradigm to investigate sensation-seeking in humans. This test probes the extent to which participants choose either to avoid or self-administer an intense tactile stimulus (mild electric stimulation) orthogonal to performance on a simple economic decision-making task. Next we investigate in a different set of participants whether this behavior is sensitive to manipulation of dopamine D2 receptors using a within-subjects, placebo-controlled, double-blind design. RESULTS: In both samples, individuals with higher self-reported sensation-seeking chose a greater proportion of mild electric stimulation-associated stimuli, even when this involved sacrifice of monetary gain. Computational modelling analysis determined that people who assigned an additional positive economic value to mild electric stimulation-associated stimuli exhibited speeding of responses when choosing these stimuli. In contrast, those who assigned a negative value exhibited slowed responses. These findings are consistent with involvement of low-level, approach-avoidance processes. Furthermore, the D2 antagonist haloperidol selectively decreased the additional economic value assigned to mild electric stimulation-associated stimuli in individuals who showed approach reactions to these stimuli under normal conditions (behavioral high-sensation seekers). CONCLUSIONS: These findings provide the first direct evidence of sensation-seeking behavior being driven by an approach-avoidance-like mechanism, modulated by dopamine, in humans. They provide a framework for investigation of psychopathologies for which extreme sensation-seeking constitutes a vulnerability factor.

Distributional reinforcement learning in prefrontal cortex.

The prefrontal cortex is crucial for learning and decision-making. Classic reinforcement learning (RL) theories center on learning the expectation of potential rewarding outcomes and explain a wealth of neural data in the prefrontal cortex. Distributional RL, on the other hand, learns the full distribution of rewarding outcomes and better explains dopamine responses. In the present study, we show that distributional RL also better explains macaque anterior cingulate cortex neuronal responses, suggesting that it is a common mechanism for reward-guided learning.

Replay and compositional computation.

Replay in the brain has been viewed as rehearsal or, more recently, as sampling from a transition model. Here, we propose a new hypothesis: that replay is able to implement a form of compositional computation where entities are assembled into relationally bound structures to derive qualitatively new knowledge. This idea builds on recent advances in neuroscience, which indicate that the hippocampus flexibly binds objects to generalizable roles and that replay strings these role-bound objects into compound statements. We suggest experiments to test our hypothesis, and we end by noting the implications for AI systems which lack the human ability to radically generalize past experience to solve new problems.

A theoretical account of cognitive effects in delay discounting.

Although delay discounting, the attenuation of the value of future rewards, is a robust finding, the mechanism of discounting is not known. We propose a potential mechanism for delay discounting such that discounting emerges from a search process that is trying to determine what rewards will be available in the future. In this theory, the delay dependence of the discounting of future expected rewards arises from three assumptions. First, that the evaluation of outcomes involves a search process. Second, that the value is assigned to an outcome proportionally to how easy it is to find. Third, that outcomes that are less delayed are typically easier for the search process to find. By relaxing this third assumption (e.g. by assuming that episodically-cued outcomes are easier to find), our model suggests that it is possible to dissociate discounting from delay. Our theory thereby explains the empirical result that discounting is slower to episodically-imagined outcomes, because these outcomes are easier for the search process to find. Additionally, the theory explains why improving cognitive resources such as working memory slows discounting, by improving searches and thereby making rewards easier to find. The three assumptions outlined here are likely to be instantiated during deliberative decision-making, but are unlikely in habitual decision-making. We model two simple implementations of this theory and show that they unify empirical results about the role of cognitive function in delay discounting, and make new neural, behavioral, and pharmacological predictions.

Replay bursts in humans coincide with activation of the default mode and parietal alpha networks.

Our brains at rest spontaneously replay recently acquired information, but how this process is orchestrated to avoid interference with ongoing cognition is an open question. Here we investigated whether replay coincided with spontaneous patterns of whole-brain activity. We found, in two separate datasets, that replay sequences were packaged into transient bursts occurring selectively during activation of the default mode network (DMN) and parietal alpha networks. These networks are believed to support inwardly oriented attention and inhibit bottom-up sensory processing and were characterized by widespread synchronized oscillations coupled to increases in high frequency power, mechanisms thought to coordinate information flow between disparate cortical areas. Our data reveal a tight correspondence between two widely studied phenomena in neural physiology and suggest that the DMN may coordinate replay bursts in a manner that minimizes interference with ongoing cognition.

Overcoming Pavlovian bias in semantic space.

Action is invigorated in the presence of reward-predicting stimuli and inhibited in the presence of punishment-predicting stimuli. Although valuable as a heuristic, this Pavlovian bias can also lead to maladaptive behaviour and is implicated in addiction. Here we explore whether Pavlovian bias can be overcome through training. Across five experiments, we find that Pavlovian bias is resistant to unlearning under most task configurations. However, we demonstrate that when subjects engage in instrumental learning in a verbal semantic space, as opposed to a motoric space, not only do they exhibit the typical Pavlovian bias, but this Pavlovian bias diminishes with training. Our results suggest that learning within the semantic space is necessary, but not sufficient, for subjects to unlearn their Pavlovian bias, and that other task features, such as gamification and spaced stimulus presentation may also be necessary. In summary, we show that Pavlovian bias, whilst robust, is susceptible to change with experience, but only under specific environmental conditions.

Dreading the pain of others? Altruistic responses to others' pain underestimate dread.

A dislike of waiting for pain, aptly termed 'dread', is so great that people will increase pain to avoid delaying it. However, despite many accounts of altruistic responses to pain in others, no previous studies have tested whether people take delay into account when attempting to ameliorate others' pain. We examined the impact of delay in 2 experiments where participants (total N = 130) specified the intensity and delay of pain either for themselves or another person. Participants were willing to increase the experimental pain of another participant to avoid delaying it, indicative of dread, though did so to a lesser extent than was the case for their own pain. We observed a similar attenuation in dread when participants chose the timing of a hypothetical painful medical treatment for a close friend or relative, but no such attenuation when participants chose for a more distant acquaintance. A model in which altruism is biased to privilege pain intensity over the dread of pain parsimoniously accounts for these findings. We refer to this underestimation of others' dread as a 'Dread Empathy Gap'.