Neural state space alignment for magnitude generalization in humans and recurrent networks.

A prerequisite for intelligent behavior is to understand how stimuli are related and to generalize this knowledge across contexts. Generalization can be challenging when relational patterns are shared across contexts but exist on different physical scales. Here, we studied neural representations in humans and recurrent neural networks performing a magnitude comparison task, for which it was advantageous to generalize concepts of "more" or "less" between contexts. Using multivariate analysis of human brain signals and of neural network hidden unit activity, we observed that both systems developed parallel neural "number lines" for each context. In both model systems, these number state spaces were aligned in a way that explicitly facilitated generalization of relational concepts (more and less). These findings suggest a previously overlooked role for neural normalization in supporting transfer of a simple form of abstract relational knowledge (magnitude) in humans and machine learning systems.

Selective Integration during Sequential Sampling in Posterior Neural Signals.

Decisions are typically made after integrating information about multiple attributes of alternatives in a choice set. Where observers are obliged to consider attributes in turn, a computational framework known as "selective integration" can capture salient biases in human choices. The model proposes that successive attributes compete for processing resources and integration is biased towards the alternative with the locally preferred attribute. Quantitative analysis shows that this model, although it discards choice-relevant information, is optimal when the observers' decisions are corrupted by noise that occurs beyond the sensory stage. Here, we used electroencephalography (EEG) to test a neural prediction of the model: that locally preferred attributes should be encoded with higher gain in neural signals over the posterior cortex. Over two sessions, human observers judged which of the two simultaneous streams of bars had the higher (or lower) average height. The selective integration model fits the data better than a rival model without bias. Single-trial analysis showed that neural signals contralateral to the preferred attribute covaried more steeply with the decision information conferred by locally preferred attributes. These findings provide neural evidence in support of selective integration, complementing existing behavioral work.

Structure learning and the posterior parietal cortex.

We propose a theory of structure learning in the primate brain. We argue that the parietal cortex is critical for learning about relations among the objects and categories that populate a visual scene. We suggest that current deep learning models exhibit poor global scene understanding because they fail to perform the relational inferences that occur in the primate dorsal stream. We review studies of neural coding in primate posterior parietal cortex (PPC), drawing the conclusion that neurons in this brain area represent potentially high-dimensional inputs on a low-dimensional manifold that encodes the relative position of objects or features in physical space, and relations among entities in abstract conceptual space. We argue that this low-dimensional code supports generalisation of relational information, even in nonspatial domains. Finally, we propose that structure learning is grounded in the actions that primates take when they reach for objects or fixate them with their eyes. We sketch a model of how this might occur in neural circuits.

Neural structure mapping in human probabilistic reward learning.

Humans can learn abstract concepts that describe invariances over relational patterns in data. One such concept, known as magnitude, allows stimuli to be compactly represented on a single dimension (i.e. on a mental line). Here, we measured representations of magnitude in humans by recording neural signals whilst they viewed symbolic numbers. During a subsequent reward-guided learning task, the neural patterns elicited by novel complex visual images reflected their payout probability in a way that suggested they were encoded onto the same mental number line, with 'bad' bandits sharing neural representation with 'small' numbers and 'good' bandits with 'large' numbers. Using neural network simulations, we provide a mechanistic model that explains our findings and shows how structural alignment can promote transfer learning. Our findings suggest that in humans, learning about reward probability is accompanied by structural alignment of value representations with neural codes for the abstract concept of magnitude.

Revisiting foraging approaches in neuroscience.

Many complex real-world decisions, such as deciding which house to buy or whether to switch jobs, involve trying to maximize reward across a sequence of choices. Optimal Foraging Theory is well suited to study these kinds of choices because it provides formal models for reward-maximization in sequential situations. In this article, we review recent insights from foraging neuroscience, behavioral ecology, and computational modelling. We find that a commonly used approach in foraging neuroscience, in which choice items are encountered at random, does not reflect the way animals direct their foraging efforts in certain real-world settings, nor does it reflect efficient reward-maximizing behavior. Based on this, we propose that task designs allowing subjects to encounter choice items strategically will further improve the ecological validity of foraging approaches used in neuroscience, as well as give rise to new behavioral and neural predictions that deepen our understanding of sequential, value-based choice.

Compulsivity Reveals a Novel Dissociation between Action and Confidence.

Confidence and actions are normally tightly interwoven-if I am sure that it is going to rain, I will take an umbrella-therefore, it is difficult to understand their interplay. Stimulated by the ego-dystonic nature of obsessive-compulsive disorder (OCD), where compulsive actions are recognized as disproportionate, we hypothesized that action and confidence might be independently updated during learning. Participants completed a predictive-inference task designed to identify how action and confidence evolve in response to surprising changes in the environment. While OCD patients (like controls) correctly updated their confidence according to changes in the environment, their actions (unlike those of controls) mostly disregarded this knowledge. Therefore, OCD patients develop an accurate, internal model of the environment but fail to use it to guide behavior. Results demonstrated a novel dissociation between confidence and action, suggesting a cognitive architecture whereby confidence estimates can accurately track the statistic of the environment independently from performance.