Adapting non-invasive human recordings along multiple task-axes shows unfolding of spontaneous and over-trained choice

Individual variability of neural computations underlying flexible decisions

The ability to flexibly switch our responses to external stimuli according to contextual information is critical for successful interactions with a complex world. Context-dependent computations are necessary across many domains1-3, yet their neural implementations remain poorly understood. Here we developed a novel behavioural task in rats to study context-dependent selection and accumulation of evidence for decision-making4-6. Under assumptions supported by both monkey and rat data, we first show mathematically that this computation can be supported by three dynamical solutions and that all networks performing the task implement a combination of these solutions. These solutions can be identified and tested directly with experimental data. We further show that existing electrophysiological and modelling data are compatible with the full variety of possible combinations of these solutions, suggesting that different individuals could use different combinations. To study variability across individual subjects, we developed automated, high-throughput methods to train rats on our task and trained many subjects using these methods. Consistent with theoretical predictions, neural and behavioural analyses revealed substantial heterogeneity across rats, despite uniformly good task performance. Our theory further predicts a specific link between behavioural and neural signatures, which was robustly supported in the data. In summary, our results provide an experimentally supported theoretical framework to analyse individual variability in biological and artificial systems that perform flexible decision-making tasks, open the door to cellular-resolution studies of individual variability in higher cognition, and provide insights into neural mechanisms of context-dependent computation more generally.

Representational spaces in orbitofrontal and ventromedial prefrontal cortex: task states, values, and beyond

The orbitofrontal cortex (OFC) and ventromedial-prefrontal cortex (vmPFC) play a key role in decision-making and encode task states in addition to expected value. We review evidence suggesting a connection between value and state representations and argue that OFC / vmPFC integrate stimulus, context, and outcome information. Comparable encoding principles emerge in late layers of deep reinforcement learning (RL) models, where single nodes exhibit similar forms of mixed-selectivity, which enables flexible readout of relevant variables by downstream neurons. Based on these lines of evidence, we suggest that outcome-maximization leads to complex representational spaces that are insufficiently characterized by linear value signals that have been the focus of most prior research on the topic. Major outstanding questions concern the role of OFC/ vmPFC in learning across tasks, in encoding of task-irrelevant aspects, and the role of hippocampus-PFC interactions.

Orthogonal neural encoding of targets and distractors supports multivariate cognitive control

The complex challenges of our mental life require us to coordinate multiple forms of neural information processing. Recent behavioural studies have found that people can coordinate multiple forms of attention, but the underlying neural control process remains obscure. We hypothesized that the brain implements multivariate control by independently monitoring feature-specific difficulty and independently prioritizing feature-specific processing. During functional MRI, participants performed a parametric conflict task that separately tags target and distractor processing. Consistent with feature-specific monitoring, univariate analyses revealed spatially segregated encoding of target and distractor difficulty in the dorsal anterior cingulate cortex. Consistent with feature-specific attentional priority, our encoding geometry analysis revealed overlapping but orthogonal representations of target and distractor coherence in the intraparietal sulcus. Coherence representations were mediated by control demands and aligned with both performance and frontoparietal activity, consistent with top-down attention. Together, these findings provide evidence for the neural geometry necessary to coordinate multivariate cognitive control.

Building compositional tasks with shared neural subspaces

Cognition is remarkably flexible; we are able to rapidly learn and perform many different tasks1. Theoretical modeling has shown artificial neural networks trained to perform multiple tasks will re-use representations2 and computational components3 across tasks. By composing tasks from these sub-components, an agent can flexibly switch between tasks and rapidly learn new tasks4. Yet, whether such compositionality is found in the brain is unknown. Here, we show the same subspaces of neural activity represent task-relevant information across multiple tasks, with each task compositionally combining these subspaces in a task-specific manner. We trained monkeys to switch between three compositionally related tasks. Neural recordings found task-relevant information about stimulus features and motor actions were represented in subspaces of neural activity that were shared across tasks. When monkeys performed a task, neural representations in the relevant shared sensory subspace were transformed to the relevant shared motor subspace. Subspaces were flexibly engaged as monkeys discovered the task in effect; their internal belief about the current task predicted the strength of representations in task-relevant subspaces. In sum, our findings suggest that the brain can flexibly perform multiple tasks by compositionally combining task-relevant neural representations across tasks.

Goal-seeking compresses neural codes for space in the human hippocampus and orbitofrontal cortex

Humans can navigate flexibly to meet their goals. Here, we asked how the neural representation of allocentric space is distorted by goal-directed behavior. Participants navigated an agent to two successive goal locations in a grid world environment comprising four interlinked rooms, with a contextual cue indicating the conditional dependence of one goal location on another. Examining the neural geometry by which room and context were encoded in fMRI signals, we found that map-like representations of the environment emerged in both hippocampus and neocortex. Cognitive maps in hippocampus and orbitofrontal cortices were compressed so that locations cued as goals were coded together in neural state space, and these distortions predicted successful learning. This effect was captured by a computational model in which current and prospective locations are jointly encoded in a place code, providing a theory of how goals warp the neural representation of space in macroscopic neural signals.

Early selection of task-relevant features through population gating

Brains can gracefully weed out irrelevant stimuli to guide behavior. This feat is believed to rely on a progressive selection of task-relevant stimuli across the cortical hierarchy, but the specific across-area interactions enabling stimulus selection are still unclear. Here, we propose that population gating, occurring within primary auditory cortex (A1) but controlled by top-down inputs from prelimbic region of medial prefrontal cortex (mPFC), can support across-area stimulus selection. Examining single-unit activity recorded while rats performed an auditory context-dependent task, we found that A1 encoded relevant and irrelevant stimuli along a common dimension of its neural space. Yet, the relevant stimulus encoding was enhanced along an extra dimension. In turn, mPFC encoded only the stimulus relevant to the ongoing context. To identify candidate mechanisms for stimulus selection within A1, we reverse-engineered low-rank RNNs trained on a similar task. Our analyses predicted that two context-modulated neural populations gated their preferred stimulus in opposite contexts, which we confirmed in further analyses of A1. Finally, we show in a two-region RNN how population gating within A1 could be controlled by top-down inputs from PFC, enabling flexible across-area communication despite fixed inter-areal connectivity.

Proactive inhibition is marked by differences in the pattern of motor cortex activity during movement preparation and execution

Successful human behavior relies on the ability to flexibly alter movements depending on the context in which they are made. One such context-dependent modulation is proactive inhibition, a type of behavioral inhibition used when anticipating the need to stop or change movements. We investigated how the motor cortex might prepare and execute movements made under different contexts. We used transcranial magnetic stimulation (TMS) in different coil orientations [postero-anterior (PA) and antero-posterior (AP) flowing currents] and pulse widths (120 and 30 µs) to probe the excitability of different inputs to corticospinal neurons while participants performed two reaction time tasks: a simple reaction time task and a stop-signal task requiring proactive inhibition. We took inspiration from state space models to assess whether the pattern of motor cortex activity changed due to proactive inhibition (PA and AP neuronal circuits represent the x and y axes of a state space upon which motor cortex activity unfolds during motor preparation and execution). We found that the rise in motor cortex excitability was delayed when proactive inhibition was required. State space visualizations showed altered patterns of motor cortex activity (combined PA₁₂₀ and AP₃₀ activity) during proactive inhibition, despite adjusting for reaction time. Overall, we show that the pattern of neural activity generated by the motor cortex during movement preparation and execution is dependent upon the context under which the movement is to be made.

NEW & NOTEWORTHY Using directional TMS, we find that the human motor cortex flexibly changes its pattern of neural activity depending on the context in which a movement is due to be made. Interestingly, this occurs despite adjusting for reaction time. We also show that state space and dynamical systems models of movement can be noninvasively visualized in humans using TMS, thereby offering a novel method to study these powerful models in humans.

Adapting non-invasive human recordings along multiple task-axes shows unfolding of spontaneous and over-trained choice

Choices rely on a transformation of sensory inputs into motor responses. Using invasive single neuron recordings, the evolution of a choice process has been tracked by projecting population neural responses into state spaces. Here, we develop an approach that allows us to recover similar trajectories on a millisecond timescale in non-invasive human recordings. We selectively suppress activity related to three task-axes, relevant and irrelevant sensory inputs and response direction, in magnetoencephalography data acquired during context-dependent choices. Recordings from premotor cortex show a progression from processing sensory input to processing the response. In contrast to previous macaque recordings, information related to choice-irrelevant features is represented more weakly than choice-relevant sensory information. To test whether this mechanistic difference between species is caused by extensive over-training common in non-human primate studies, we trained humans on >20,000 trials of the task. Choice-irrelevant features were still weaker than relevant features in premotor cortex after over-training.