Geometry of sequence working memory in macaque prefrontal cortex

Timescales of learning in prefrontal cortex

The lateral prefrontal cortex (PFC) in humans and other primates is critical for immediate, goal-directed behaviour and working memory, which are classically considered distinct from the cognitive and neural circuits that support long-term learning and memory. Over the past few years, a reconsideration of this textbook perspective has emerged, in that different timescales of memory-guided behaviour are in constant interaction during the pursuit of immediate goals. Here, we will first detail how neural activity related to the shortest timescales of goal-directed behaviour (which requires maintenance of current states and goals in working memory) is sculpted by long-term knowledge and learning - that is, how the past informs present behaviour. Then, we will outline how learning across different timescales (from seconds to years) drives plasticity in the primate lateral PFC, from single neuron firing rates to mesoscale neuroimaging activity patterns. Finally, we will review how, over days and months of learning, dense local and long-range connectivity patterns in PFC facilitate longer-lasting changes in population activity by changing synaptic weights and recruiting additional neural resources to inform future behaviour. Our Review sheds light on how the machinery of plasticity in PFC circuits facilitates the integration of learned experiences across time to best guide adaptive behaviour.

Dissociable functional responses along the posterior-anterior gradient of the frontal and parietal cortices revealed by parametric working memory and training

While the storage capacity is limited, accumulating studies have indicated that working memory (WM) can be improved by cognitive training. However, understanding how exactly the brain copes with limited WM capacity and how cognitive training optimizes the brain remains inconclusive. Given the hierarchical functional organization of WM, we hypothesized that the activation profiles along the posterior-anterior gradient of the frontal and parietal cortices characterize WM load and training effects. To test this hypothesis, we recruited 51 healthy volunteers and adopted a parametric WM paradigm and training method. In contrast to exclusively strengthening the activation of posterior areas, a broader range of activation concurrently occurred in the anterior areas to cope with increased memory load for all subjects at baseline. Moreover, there was an imbalance in the responses of the posterior and anterior areas to the same increment of 1 item at different load levels. Although a general decrease in activation after adaptive training, the changes in the posterior and anterior areas were distinct at different memory loads. Particularly, we found that the activation gradient between the posterior and anterior areas was significantly increased at load 4-back after adaptive training, and the changes were correlated with improvement in WM performance. Together, our results demonstrate a shift in the predominant role of posterior and anterior areas in the frontal and parietal cortices when approaching WM capacity limits. Additionally, the training-induced performance improvement likely benefits from the elevated neural efficiency reflected in the increased activation gradient between the posterior and anterior areas.

Flexible control of sequence working memory in the macaque frontal cortex

To memorize a sequence, one must serially bind each item to its rank order. How the brain controls a given input to bind its associated order in sequence working memory (SWM) remains unexplored. Here, we investigated the neural representations underlying SWM control using electrophysiological recordings in the frontal cortex of macaque monkeys performing forward and backward SWM tasks. Separate and generalizable low-dimensional subspaces for sensory and memory information were found within the same frontal circuitry, and SWM control was reflected in these neural subspaces' organized dynamics. Each item at each rank was sequentially entered into a common sensory subspace and, depending on forward or backward task requirement, flexibly and timely sent into rank-selective SWM subspaces. Neural activity in these SWM subspaces faithfully predicted the recalled item and order information in single error trials. Thus, compositional neural population codes with well-orchestrated dynamics in frontal cortex support the flexible control of SWM.

Prefrontal neuronal dynamics in the absence of task execution

Prefrontal cortical activity represents stimuli in working memory tasks in a low-dimensional manifold that transforms over the course of a trial. Such transformations reflect specific cognitive operations, so that, for example, the rotation of stimulus representations is thought to reduce interference by distractor stimuli. Here we show that rotations occur in the low-dimensional activity space of prefrontal neurons in naïve male monkeys (Macaca mulatta), while passively viewing familiar stimuli. Moreover, some aspects of these rotations remain remarkably unchanged after training to perform working memory tasks. Significant training effects are still present in population dynamics, which further distinguish correct and error trials during task execution. Our results reveal automatic functions of prefrontal neural circuits allow transformations that may aid cognitive flexibility.

Neural representational geometries reflect behavioral differences in monkeys and recurrent neural networks

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 male 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 are associated with differences in the reaction times, revealing behavioral differences we were unaware of. All these analyses suggest 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.

Orthogonality of sensory and contextual categorical dynamics embedded in a continuum of responses from the second somatosensory cortex

How does the brain simultaneously process signals that bring complementary information, like raw sensory signals and their transformed counterparts, without any disruptive interference? Contemporary research underscores the brain's adeptness in using decorrelated responses to reduce such interference. Both neurophysiological findings and artificial neural networks support the notion of orthogonal representation for signal differentiation and parallel processing. Yet, where, and how raw sensory signals are transformed into more abstract representations remains unclear. Using a temporal pattern discrimination task in trained monkeys, we revealed that the second somatosensory cortex (S2) efficiently segregates faithful and transformed neural responses into orthogonal subspaces. Importantly, S2 population encoding for transformed signals, but not for faithful ones, disappeared during a nondemanding version of this task, which suggests that signal transformation and their decoding from downstream areas are only active on-demand. A mechanistic computation model points to gain modulation as a possible biological mechanism for the observed context-dependent computation. Furthermore, individual neural activities that underlie the orthogonal population representations exhibited a continuum of responses, with no well-determined clusters. These findings advocate that the brain, while employing a continuum of heterogeneous neural responses, splits population signals into orthogonal subspaces in a context-dependent fashion to enhance robustness, performance, and improve coding efficiency.

Cost-benefit Tradeoff Mediates the Rule- to Memory-based Processing Transition during Practice

Practice not only improves task performance, but also changes task execution from rule- to memory-based processing by incorporating experiences from practice. However, how and when this change occurs is unclear. We tested the hypothesis that strategy transition in task learning results from cost-benefit analysis. Participants learned two task sequences and were then queried about the task type at a cued sequence and position. Behavioral improvement with practice can be accounted for by a computational model implementing cost-benefit analysis. Model-predicted strategy transition points are related to behavioral slowing and changes in fMRI activation patterns in the dorsolateral prefrontal cortex. Strategy transition is also related to increased pattern separation in the ventromedial prefrontal cortex. The cost-benefit analysis model outperforms alternative models (e.g., both strategies racing for being expressed in behavior) in accounting for empirical data. These findings support cost-benefit analysis as a mechanism of practice-induced strategy shift.

Representations and generalization in artificial and brain neural networks

Humans and animals excel at generalizing from limited data, a capability yet to be fully replicated in artificial intelligence. This perspective investigates generalization in biological and artificial deep neural networks (DNNs), in both in-distribution and out-of-distribution contexts. We introduce two hypotheses: First, the geometric properties of the neural manifolds associated with discrete cognitive entities, such as objects, words, and concepts, are powerful order parameters. They link the neural substrate to the generalization capabilities and provide a unified methodology bridging gaps between neuroscience, machine learning, and cognitive science. We overview recent progress in studying the geometry of neural manifolds, particularly in visual object recognition, and discuss theories connecting manifold dimension and radius to generalization capacity. Second, we suggest that the theory of learning in wide DNNs, especially in the thermodynamic limit, provides mechanistic insights into the learning processes generating desired neural representational geometries and generalization. This includes the role of weight norm regularization, network architecture, and hyper-parameters. We will explore recent advances in this theory and ongoing challenges. We also discuss the dynamics of learning and its relevance to the issue of representational drift in the brain.

Computational role of structure in neural activity and connectivity

One major challenge of neuroscience is identifying structure in seemingly disorganized neural activity. Different types of structure have different computational implications that can help neuroscientists understand the functional role of a particular brain area. Here, we outline a unified approach to characterize structure by inspecting the representational geometry and the modularity properties of the recorded activity and show that a similar approach can also reveal structure in connectivity. We start by setting up a general framework for determining geometry and modularity in activity and connectivity and relating these properties with computations performed by the network. We then use this framework to review the types of structure found in recent studies of model networks performing three classes of computations.

Beta: bursts of cognition

Beta oscillations are linked to the control of goal-directed processing of sensory information and the timing of motor output. Recent evidence demonstrates they are not sustained but organized into intermittent high-power bursts mediating timely functional inhibition. This implies there is a considerable moment-to-moment variation in the neural dynamics supporting cognition. Beta bursts thus offer new opportunities for studying how sensory inputs are selectively processed, reshaped by inhibitory cognitive operations and ultimately result in motor actions. Recent method advances reveal diversity in beta bursts that provide deeper insights into their function and the underlying neural circuit activity motifs. We propose that brain-wide, spatiotemporal patterns of beta bursting reflect various cognitive operations and that their dynamics reveal nonlinear aspects of cortical processing.

Neuronal activation sequences in lateral prefrontal cortex encode visuospatial working memory during virtual navigation

Working memory (WM) is the ability to maintain and manipulate information 'in mind'. The neural codes underlying WM have been a matter of debate. We simultaneously recorded the activity of hundreds of neurons in the lateral prefrontal cortex of male macaque monkeys during a visuospatial WM task that required navigation in a virtual 3D environment. Here, we demonstrate distinct neuronal activation sequences (NASs) that encode remembered target locations in the virtual environment. This NAS code outperformed the persistent firing code for remembered locations during the virtual reality task, but not during a classical WM task using stationary stimuli and constraining eye movements. Finally, blocking NMDA receptors using low doses of ketamine deteriorated the NAS code and behavioral performance selectively during the WM task. These results reveal the versatility and adaptability of neural codes supporting working memory function in the primate lateral prefrontal cortex.

Jointly looking to the past and the future in visual working memory

Working memory enables us to bridge past sensory information to upcoming future behaviour. Accordingly, by its very nature, working memory is concerned with two components: the past and the future. Yet, in conventional laboratory tasks, these two components are often conflated, such as when sensory information in working memory is encoded and tested at the same location. We developed a task in which we dissociated the past (encoded location) and future (to-be-tested location) attributes of visual contents in working memory. This enabled us to independently track the utilisation of past and future memory attributes through gaze, as observed during mnemonic selection. Our results reveal the joint consideration of past and future locations. This was prevalent even at the single-trial level of individual saccades that were jointly biased to the past and future. This uncovers the rich nature of working memory representations, whereby both past and future memory attributes are retained and can be accessed together when memory contents become relevant for behaviour.

Anti-Hebbian plasticity drives sequence learning in striatum

Spatio-temporal activity patterns have been observed in a variety of brain areas in spontaneous activity, prior to or during action, or in response to stimuli. Biological mechanisms endowing neurons with the ability to distinguish between different sequences remain largely unknown. Learning sequences of spikes raises multiple challenges, such as maintaining in memory spike history and discriminating partially overlapping sequences. Here, we show that anti-Hebbian spike-timing dependent plasticity (STDP), as observed at cortico-striatal synapses, can naturally lead to learning spike sequences. We design a spiking model of the striatal output neuron receiving spike patterns defined as sequential input from a fixed set of cortical neurons. We use a simple synaptic plasticity rule that combines anti-Hebbian STDP and non-associative potentiation for a subset of the presented patterns called rewarded patterns. We study the ability of striatal output neurons to discriminate rewarded from non-rewarded patterns by firing only after the presentation of a rewarded pattern. In particular, we show that two biological properties of striatal networks, spiking latency and collateral inhibition, contribute to an increase in accuracy, by allowing a better discrimination of partially overlapping sequences. These results suggest that anti-Hebbian STDP may serve as a biological substrate for learning sequences of spikes.

Learning leaves a memory trace in motor cortex

How are we able to learn new behaviors without disrupting previously learned ones? To understand how the brain achieves this, we used a brain-computer interface (BCI) learning paradigm, which enables us to detect the presence of a memory of one behavior while performing another. We found that learning to use a new BCI map altered the neural activity that monkeys produced when they returned to using a familiar BCI map in a way that was specific to the learning experience. That is, learning left a "memory trace" in the primary motor cortex. This memory trace coexisted with proficient performance under the familiar map, primarily by altering neural activity in dimensions that did not impact behavior. Forming memory traces might be how the brain is able to provide for the joint learning of multiple behaviors without interference.

Neural dynamic foundations of a theory of higher cognition: the case of grounding nested phrases

Because cognitive competences emerge in evolution and development from the sensory-motor domain, we seek a neural process account for higher cognition in which all representations are necessarily grounded in perception and action. The challenge is to understand how hallmarks of higher cognition, productivity, systematicity, and compositionality, may emerge from such a bottom-up approach. To address this challenge, we present key ideas from Dynamic Field Theory which postulates that neural populations are organized by recurrent connectivity to create stable localist representations. Dynamic instabilities enable the autonomous generation of sequences of mental states. The capacity to apply neural circuitry across broad sets of inputs that emulates the function call postulated in symbolic computation emerges through coordinate transforms implemented in neural gain fields. We show how binding localist neural representations through a shared index dimension enables conceptual structure, in which the interdependence among components of a representation is flexibly expressed. We demonstrate these principles in a neural dynamic architecture that represents and perceptually grounds nested relational and action phrases. Sequences of neural processing steps are generated autonomously to attentionally select the referenced objects and events in a manner that is sensitive to their interdependencies. This solves the problem of 2 and the massive binding problem in expressions such as "the small tree that is to the left of the lake which is to the left of the large tree". We extend earlier work by incorporating new types of grammatical constructions and a larger vocabulary. We discuss the DFT framework relative to other neural process accounts of higher cognition and assess the scope and challenges of such neural theories.

Spontaneously emerging internal models of visual sequences combine abstract and event-specific information in the prefrontal cortex

When exposed to sensory sequences, do macaque monkeys spontaneously form abstract internal models that generalize to novel experiences? Here, we show that neuronal populations in macaque ventrolateral prefrontal cortex jointly encode visual sequences by separate codes for the specific pictures presented and for their abstract sequential structure. We recorded prefrontal neurons while macaque monkeys passively viewed visual sequences and sequence mismatches in the local-global paradigm. Even without any overt task or response requirements, prefrontal populations spontaneously form representations of sequence structure, serial order, and image identity within distinct but superimposed neuronal subspaces. Representations of sequence structure rapidly update following single exposure to a mismatch sequence, while distinct populations represent mismatches for sequences of different complexity. Finally, those representations generalize across sequences following the same repetition structure but comprising different images. These results suggest that prefrontal populations spontaneously encode rich internal models of visual sequences reflecting both content-specific and abstract information.

Tracking the neural codes for words and phrases during semantic composition, working-memory storage, and retrieval

The ability to compose successive words into a meaningful phrase is a characteristic feature of human cognition, yet its neural mechanisms remain incompletely understood. Here, we analyze the cortical mechanisms of semantic composition using magnetoencephalography (MEG) while participants read one-word, two-word, and five-word noun phrases and compared them with a subsequent image. Decoding of MEG signals revealed three processing stages. During phrase comprehension, the representation of individual words was sustained for a variable duration depending on phrasal context. During the delay period, the word code was replaced by a working-memory code whose activation increased with semantic complexity. Finally, the speed and accuracy of retrieval depended on semantic complexity and was faster for surface than for deep semantic properties. In conclusion, we propose that the brain initially encodes phrases using factorized dimensions for successive words but later compresses them in working memory and requires a period of decompression to access them.

Distinct neural signatures underlying information maintenance and manipulation in working memory

Previous working memory research has demonstrated robust stimulus representations during memory maintenance in both voltage and alpha-band activity in electroencephalography. However, the exact functions of these 2 neural signatures have remained controversial. Here we systematically investigated their respective contributions to memory manipulation. Human participants either maintained a previously seen spatial location, or manipulated the location following a mental rotation cue over a delay. Using multivariate decoding, we observed robust location representations in low-frequency voltage and alpha-band oscillatory activity with distinct spatiotemporal dynamics: location representations were most evident in posterior channels in alpha-band activity, but were most prominent in the more anterior, central channels in voltage signals. Moreover, the temporal emergence of manipulated representation in central voltage preceded that in posterior alpha-band activity, suggesting that voltage might carry stimulus-specific source signals originated internally from anterior cortex, whereas alpha-band activity might reflect feedback signals in posterior cortex received from higher-order cortex. Lastly, while location representations in both signals were coded in a low-dimensional neural subspace, location representation in central voltage was higher-dimensional and underwent a representational transformation that exclusively predicted memory behavior. Together, these results highlight the crucial role of central voltage in working memory, and support functional distinctions between voltage and alpha-band activity.

Mesoscopic calcium imaging in a head-unrestrained male non-human primate using a lensless microscope

Mesoscopic calcium imaging enables studies of cell-type specific neural activity over large areas. A growing body of literature suggests that neural activity can be different when animals are free to move compared to when they are restrained. Unfortunately, existing systems for imaging calcium dynamics over large areas in non-human primates (NHPs) are table-top devices that require restraint of the animal's head. Here, we demonstrate an imaging device capable of imaging mesoscale calcium activity in a head-unrestrained male non-human primate. We successfully miniaturize our system by replacing lenses with an optical mask and computational algorithms. The resulting lensless microscope can fit comfortably on an NHP, allowing its head to move freely while imaging. We are able to measure orientation columns maps over a 20 mm2 field-of-view in a head-unrestrained macaque. Our work establishes mesoscopic imaging using a lensless microscope as a powerful approach for studying neural activity under more naturalistic conditions.

Theta Oscillations Support Prefrontal-hippocampal Interactions in Sequential Working Memory

The prefrontal cortex and hippocampus may support sequential working memory beyond episodic memory and spatial navigation. This stereoelectroencephalography (SEEG) study investigated how the dorsolateral prefrontal cortex (DLPFC) interacts with the hippocampus in the online processing of sequential information. Twenty patients with epilepsy (eight women, age 27.6 ± 8.2 years) completed a line ordering task with SEEG recordings over the DLPFC and the hippocampus. Participants showed longer thinking times and more recall errors when asked to arrange random lines clockwise (random trials) than to maintain ordered lines (ordered trials) before recalling the orientation of a particular line. First, the ordering-related increase in thinking time and recall error was associated with a transient theta power increase in the hippocampus and a sustained theta power increase in the DLPFC (3-10 Hz). In particular, the hippocampal theta power increase correlated with the memory precision of line orientation. Second, theta phase coherences between the DLPFC and hippocampus were enhanced for ordering, especially for more precisely memorized lines. Third, the theta band DLPFC → hippocampus influence was selectively enhanced for ordering, especially for more precisely memorized lines. This study suggests that theta oscillations may support DLPFC-hippocampal interactions in the online processing of sequential information.

Temporal and Spatial Information Elicit Different Power and Connectivity Profiles during Working Memory Maintenance

Working memory (WM) is the cognitive ability to store and manipulate information necessary for ongoing tasks. Although frontoparietal areas are involved in the retention of visually presented information, oscillatory neural activity differs for temporal and spatial WM processing. In this study, we corroborated previous findings describing the modulation of neural oscillations and expanded our investigation to the network organization underlying the cognitive processing of temporal and spatial information. We utilized MEG recordings during a Sternberg visual WM task. The spectral oscillatory activity in the maintenance phase revealed increased frontal theta (4-8 Hz) and parietal beta (13-30 Hz) in the temporal condition. Source level coherence analysis delineated the prominent role of parietal areas in all frequency bands during the maintenance of temporal information, whereas frontal and central areas showed major contributions in theta and beta ranges during the maintenance of spatial information. Our study revealed distinct spectral profiles of neural oscillations for separate cognitive subdomains of WM processing. The delineation of specific functional networks might have important implications for clinical applications, enabling the development of stimulation protocols targeting cognitive disabilities associated with WM impairments.

A construct-first approach to consciousness science

We propose a new approach to consciousness science that instead of comparing complex theoretical positions deconstructs existing theories, takes their central assumptions while disregarding their auxiliary hypotheses, and focuses its investigations on the main constructs that these central assumptions rely on (like global workspace, recurrent processing, metarepresentation). Studying how these main constructs are anchored in lower-level constructs characterizing underlying neural processing will not just offer an alternative to theory comparisons but will also take us one step closer to empirical resolutions. Moreover, exploring the compatibility and possible combinations of the lower-level constructs will allow for new theoretical syntheses. This construct-first approach will improve our ability to understand the commitments of existing theories and pave the way for moving beyond them.

Signatures of task learning in neural representations

While neural plasticity has long been studied as the basis of learning, the growth of large-scale neural recording techniques provides a unique opportunity to study how learning-induced activity changes are coordinated across neurons within the same circuit. These distributed changes can be understood through an evolution of the geometry of neural manifolds and latent dynamics underlying new computations. In parallel, studies of multi-task and continual learning in artificial neural networks hint at a tradeoff between non-interference and compositionality as guiding principles to understand how neural circuits flexibly support multiple behaviors. In this review, we highlight recent findings from both biological and artificial circuits that together form a new framework for understanding task learning at the population level.

Functional alterations of the prefrontal circuit underlying cognitive aging in mice

Executive function is susceptible to aging. How aging impacts the circuit-level computations underlying executive function remains unclear. Using calcium imaging and optogenetic manipulation during memory-guided behavior, we show that working-memory coding and the relevant recurrent connectivity in the mouse medial prefrontal cortex (mPFC) are altered as early as middle age. Population activity in the young adult mPFC exhibits dissociable yet overlapping patterns between tactile and auditory modalities, enabling crossmodal memory coding concurrent with modality-dependent coding. In middle age, however, crossmodal coding remarkably diminishes while modality-dependent coding persists, and both types of coding decay in advanced age. Resting-state functional connectivity, especially among memory-coding neurons, decreases already in middle age, suggesting deteriorated recurrent circuits for memory maintenance. Optogenetic inactivation reveals that the middle-aged mPFC exhibits heightened vulnerability to perturbations. These findings elucidate functional alterations of the prefrontal circuit that unfold in middle age and deteriorate further as a hallmark of cognitive aging.

Rhesus monkeys exhibiting spontaneous ritualistic behaviors resembling obsessive-compulsive disorder

Obsessive-compulsive disorder (OCD) is a chronic and debilitating psychiatric disorder that affects ∼2%-3% of the population globally. Studying spontaneous OCD-like behaviors in non-human primates may improve our understanding of the disorder. In large rhesus monkey colonies, we found 10 monkeys spontaneously exhibiting persistent sequential motor behaviors (SMBs) in individual-specific sequences that were repetitive, time-consuming and stable over prolonged periods. Genetic analysis revealed severely damaging mutations in genes associated with OCD risk in humans. Brain imaging showed that monkeys with SMBs had larger gray matter (GM) volumes in the left caudate nucleus and lower fractional anisotropy of the corpus callosum. The GM volume of the left caudate nucleus correlated positively with the daily duration of SMBs. Notably, exposure to a stressor (human presence) significantly increased SMBs. In addition, fluoxetine, a serotonergic medication commonly used for OCD, decreased SMBs in these monkeys. These findings provide a novel foundation for developing better understanding and treatment of OCD.

Oligodendrocyte dynamics dictate cognitive performance outcomes of working memory training in mice

Previous work has shown that motor skill learning stimulates and requires generation of myelinating oligodendrocytes (OLs) from their precursor cells (OLPs) in the brains of adult mice. In the present study we ask whether OL production is also required for non-motor learning and cognition, using T-maze and radial-arm-maze tasks that tax spatial working memory. We find that maze training stimulates OLP proliferation and OL production in the medial prefrontal cortex (mPFC), anterior corpus callosum (genu), dorsal thalamus and hippocampal formation of adult male mice; myelin sheath formation is also stimulated in the genu. Genetic blockade of OL differentiation and neo-myelination in Myrf conditional-knockout mice strongly impairs training-induced improvements in maze performance. We find a strong positive correlation between the performance of individual wild type mice and the scale of OLP proliferation and OL generation during training, but not with the number or intensity of c-Fos+ neurons in their mPFC, underscoring the important role played by OL lineage cells in cognitive processing.

Pseudo-linear Summation explains Neural Geometry of Multi-finger Movements in Human Premotor Cortex

How does the motor cortex combine simple movements (such as single finger flexion/extension) into complex movements (such hand gestures or playing piano)? Motor cortical activity was recorded using intracortical multi-electrode arrays in two people with tetraplegia as they attempted single, pairwise and higher order finger movements. Neural activity for simultaneous movements was largely aligned with linear summation of corresponding single finger movement activities, with two violations. First, the neural activity was normalized, preventing a large magnitude with an increasing number of moving fingers. Second, the neural tuning direction of weakly represented fingers (e.g. middle) changed significantly as a result of the movement of other fingers. These deviations from linearity resulted in non-linear methods outperforming linear methods for neural decoding. Overall, simultaneous finger movements are thus represented by the combination of individual finger movements by pseudo-linear summation.

A recurrent neural network model of prefrontal brain activity during a working memory task

When multiple items are held in short-term memory, cues that retrospectively prioritise one item over another (retro-cues) can facilitate subsequent recall. However, the neural and computational underpinnings of this effect are poorly understood. One recent study recorded neural signals in the macaque lateral prefrontal cortex (LPFC) during a retro-cueing task, contrasting delay-period activity before (pre-cue) and after (post-cue) retrocue onset. They reported that in the pre-cue delay, the individual stimuli were maintained in independent subspaces of neural population activity, whereas in the post-cue delay, the prioritised items were rotated into a common subspace, potentially allowing a common readout mechanism. To understand how such representational transitions can be learnt through error minimisation, we trained recurrent neural networks (RNNs) with supervision to perform an equivalent cued-recall task. RNNs were presented with two inputs denoting conjunctive colour-location stimuli, followed by a pre-cue memory delay, a location retrocue, and a post-cue delay. We found that the orthogonal-to-parallel geometry transformation observed in the macaque LPFC emerged naturally in RNNs trained to perform the task. Interestingly, the parallel geometry only developed when the cued information was required to be maintained in short-term memory for several cycles before readout, suggesting that it might confer robustness during maintenance. We extend these findings by analysing the learning dynamics and connectivity patterns of the RNNs, as well as the behaviour of models trained with probabilistic cues, allowing us to make predictions for future studies. Overall, our findings are consistent with recent theoretical accounts which propose that retrocues transform the prioritised memory items into a prospective, action-oriented format.

Neural population dynamics of human working memory

The activity of neurons in macaque prefrontal cortex (PFC) persists during working memory (WM) delays, providing a mechanism for memory.1,2,3,4,5,6,7,8,9,10,11 Although theory,11,12 including formal network models,13,14 assumes that WM codes are stable over time, PFC neurons exhibit dynamics inconsistent with these assumptions.15,16,17,18,19 Recently, multivariate reanalyses revealed the coexistence of both stable and dynamic WM codes in macaque PFC.20,21,22,23 Human EEG studies also suggest that WM might contain dynamics.24,25 Nonetheless, how WM dynamics vary across the cortical hierarchy and which factors drive dynamics remain unknown. To elucidate WM dynamics in humans, we decoded WM content from fMRI responses across multiple cortical visual field maps.26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48 We found coexisting stable and dynamic neural representations of WM during a memory-guided saccade task. Geometric analyses of neural subspaces revealed that early visual cortex exhibited stronger dynamics than high-level visual and frontoparietal cortex. Leveraging models of population receptive fields, we visualized and made the neural dynamics interpretable. We found that during WM delays, V1 population initially encoded a narrowly tuned bump of activation centered on the peripheral memory target. Remarkably, this bump then spread inward toward foveal locations, forming a vector along the trajectory of the forthcoming memory-guided saccade. In other words, the neural code transformed into an abstraction of the stimulus more proximal to memory-guided behavior. Therefore, theories of WM must consider both sensory features and their task-relevant abstractions because changes in the format of memoranda naturally drive neural dynamics.

Temporal logic circuits implementation using a dual cross-inhibition mechanism based on DNA strand displacement

Molecular circuits crafted from DNA molecules harness the inherent programmability and biocompatibility of DNA to intelligently steer molecular machines in the execution of microscopic tasks. In comparison to combinational circuits, DNA-based temporal circuits boast supplementary capabilities, allowing them to proficiently handle the omnipresent temporal information within biochemical systems and life sciences. However, the lack of temporal mechanisms and components proficient in comprehending and processing temporal information presents challenges in advancing DNA circuits that excel in complex tasks requiring temporal control and time perception. In this study, we engineered temporal logic circuits through the design and implementation of a dual cross-inhibition mechanism, which enables the acceptance and processing of temporal information, serving as a fundamental building block for constructing temporal circuits. By incorporating the dual cross-inhibition mechanism, the temporal logic gates are endowed with cascading capabilities, significantly enhancing the inhibitory effect compared to a cross-inhibitor. Furthermore, we have introduced the annihilation mechanism into the circuit to further augment the inhibition effect. As a result, the circuit demonstrates sensitive time response characteristics, leading to a fundamental improvement in circuit performance. This architecture provides a means to efficiently process temporal signals in DNA strand displacement circuits. We anticipate that our findings will contribute to the design of complex temporal logic circuits and the advancement of molecular programming.

Abstraction of Reward Context Facilitates Relative Reward Coding in Neural Populations of the Macaque Anterior Cingulate Cortex

The anterior cingulate cortex (ACC) is believed to be involved in many cognitive processes, including linking goals to actions and tracking decision-relevant contextual information. ACC neurons robustly encode expected outcomes, but how this relates to putative functions of ACC remains unknown. Here, we approach this question from the perspective of population codes by analyzing neural spiking data in the ventral and dorsal banks of the ACC in two male monkeys trained to perform a stimulus-motor mapping task to earn rewards or avoid losses. We found that neural populations favor a low dimensional representational geometry that emphasizes the valence of potential outcomes while also facilitating the independent, abstract representation of multiple task-relevant variables. Valence encoding persisted throughout the trial, and realized outcomes were primarily encoded in a relative sense, such that cue valence acted as a context for outcome encoding. This suggests that the population coding we observe could be a mechanism that allows feedback to be interpreted in a context-dependent manner. Together, our results point to a prominent role for ACC in context setting and relative interpretation of outcomes, facilitated by abstract, or untangled, representations of task variables.SIGNIFICANCE STATEMENT The ability to interpret events in light of the current context is a critical facet of higher-order cognition. The ACC is suggested to be important for tracking contextual information, whereas alternate views hold that its function is more related to the motor system and linking goals to appropriate actions. We evaluated these possibilities by analyzing geometric properties of neural population activity in monkey ACC when contexts were determined by the valence of potential outcomes and found that this information was represented as a dominant, abstract concept. Ensuing outcomes were then coded relative to these contexts, suggesting an important role for these representations in context-dependent evaluation. Such mechanisms may be critical for the abstract reasoning and generalization characteristic of biological intelligence.

Action space restructures visual working memory in prefrontal cortex

Visual working memory enables flexible behavior by decoupling sensory stimuli from behavioral actions. While previous studies have predominantly focused on the storage component of working memory, the role of future actions in shaping working memory remains unknown. To answer this question, we used two working memory tasks that allowed the dissociation of sensory and action components of working memory. We measured behavioral performance and neuronal activity in the macaque prefrontal cortex area, frontal eye fields. We show that the action space reshapes working memory, as evidenced by distinct patterns of memory tuning and attentional orienting between the two tasks. Notably, neuronal activity during the working memory period predicted future behavior and exhibited mixed selectivity in relation to the sensory space but linear selectivity relative to the action space. This linear selectivity was achieved through the rapid transformation from sensory to action space and was subsequently maintained as a stable cross-temporal population activity pattern. Combined, we provide direct physiological evidence of the action-oriented nature of frontal eye field neurons during memory tasks and demonstrate that the anticipation of behavioral outcomes plays a significant role in transforming and maintaining the contents of visual working memory.

Dimensionality and Ramping: Signatures of Sentence Integration in the Dynamics of Brains and Deep Language Models

A sentence is more than the sum of its words: its meaning depends on how they combine with one another. The brain mechanisms underlying such semantic composition remain poorly understood. To shed light on the neural vector code underlying semantic composition, we introduce two hypotheses: (1) the intrinsic dimensionality of the space of neural representations should increase as a sentence unfolds, paralleling the growing complexity of its semantic representation; and (2) this progressive integration should be reflected in ramping and sentence-final signals. To test these predictions, we designed a dataset of closely matched normal and jabberwocky sentences (composed of meaningless pseudo words) and displayed them to deep language models and to 11 human participants (5 men and 6 women) monitored with simultaneous MEG and intracranial EEG. In both deep language models and electrophysiological data, we found that representational dimensionality was higher for meaningful sentences than jabberwocky. Furthermore, multivariate decoding of normal versus jabberwocky confirmed three dynamic patterns: (1) a phasic pattern following each word, peaking in temporal and parietal areas; (2) a ramping pattern, characteristic of bilateral inferior and middle frontal gyri; and (3) a sentence-final pattern in left superior frontal gyrus and right orbitofrontal cortex. These results provide a first glimpse into the neural geometry of semantic integration and constrain the search for a neural code of linguistic composition.SIGNIFICANCE STATEMENT Starting from general linguistic concepts, we make two sets of predictions in neural signals evoked by reading multiword sentences. First, the intrinsic dimensionality of the representation should grow with additional meaningful words. Second, the neural dynamics should exhibit signatures of encoding, maintaining, and resolving semantic composition. We successfully validated these hypotheses in deep neural language models, artificial neural networks trained on text and performing very well on many natural language processing tasks. Then, using a unique combination of MEG and intracranial electrodes, we recorded high-resolution brain data from human participants while they read a controlled set of sentences. Time-resolved dimensionality analysis showed increasing dimensionality with meaning, and multivariate decoding allowed us to isolate the three dynamical patterns we had hypothesized.

Advances in AAV technology for delivering genetically encoded cargo to the nonhuman primate nervous system

Modern neuroscience approaches including optogenetics, calcium imaging, and other genetic manipulations have facilitated our ability to dissect specific circuits in rodent models to study their role in neurological disease. These approaches regularly use viral vectors to deliver genetic cargo (e.g., opsins) to specific tissues and genetically-engineered rodents to achieve cell-type specificity. However, the translatability of these rodent models, cross-species validation of identified targets, and translational efficacy of potential therapeutics in larger animal models like nonhuman primates remains difficult due to the lack of efficient primate viral vectors. A refined understanding of the nonhuman primate nervous system promises to deliver insights that can guide the development of treatments for neurological and neurodegenerative diseases. Here, we outline recent advances in the development of adeno-associated viral vectors for optimized use in nonhuman primates. These tools promise to help open new avenues for study in translational neuroscience and further our understanding of the primate brain.

Comparing representations and computations in single neurons versus neural networks

Single-neuron-level explanations have been the gold standard in neuroscience for decades. Recently, however, neural-network-level explanations have become increasingly popular. This increase in popularity is driven by the fact that the analysis of neural networks can solve problems that cannot be addressed by analyzing neurons independently. In this opinion article, I argue that while both frameworks employ the same general logic to link physical and mental phenomena, in many cases the neural network framework provides better explanatory objects to understand representations and computations related to mental phenomena. I discuss what constitutes a mechanistic explanation in neural systems, provide examples, and conclude by highlighting a number of the challenges and considerations associated with the use of analyses of neural networks to study brain function.

How our understanding of memory replay evolves

Memory reactivations and replay, widely reported in the hippocampus and cortex across species, have been implicated in memory consolidation, planning, and spatial and skill learning. Technological advances in electrophysiology, calcium imaging, and human neuroimaging techniques have enabled neuroscientists to measure large-scale neural activity with increasing spatiotemporal resolution and have provided opportunities for developing robust analytic methods to identify memory replay. In this article, we first review a large body of historically important and representative memory replay studies from the animal and human literature. We then discuss our current understanding of memory replay functions in learning, planning, and memory consolidation and further discuss the progress in computational modeling that has contributed to these improvements. Next, we review past and present analytic methods for replay analyses and discuss their limitations and challenges. Finally, looking ahead, we discuss some promising analytic methods for detecting nonstereotypical, behaviorally nondecodable structures from large-scale neural recordings. We argue that seamless integration of multisite recordings, real-time replay decoding, and closed-loop manipulation experiments will be essential for delineating the role of memory replay in a wide range of cognitive and motor functions.

Continual task learning in natural and artificial agents

How do humans and other animals learn new tasks? A wave of brain recording studies has investigated how neural representations change during task learning, with a focus on how tasks can be acquired and coded in ways that minimise mutual interference. We review recent work that has explored the geometry and dimensionality of neural task representations in neocortex, and computational models that have exploited these findings to understand how the brain may partition knowledge between tasks. We discuss how ideas from machine learning, including those that combine supervised and unsupervised learning, are helping neuroscientists understand how natural tasks are learned and coded in biological brains.

Modelling continual learning in humans with Hebbian context gating and exponentially decaying task signals

Humans can learn several tasks in succession with minimal mutual interference but perform more poorly when trained on multiple tasks at once. The opposite is true for standard deep neural networks. Here, we propose novel computational constraints for artificial neural networks, inspired by earlier work on gating in the primate prefrontal cortex, that capture the cost of interleaved training and allow the network to learn two tasks in sequence without forgetting. We augment standard stochastic gradient descent with two algorithmic motifs, so-called "sluggish" task units and a Hebbian training step that strengthens connections between task units and hidden units that encode task-relevant information. We found that the "sluggish" units introduce a switch-cost during training, which biases representations under interleaved training towards a joint representation that ignores the contextual cue, while the Hebbian step promotes the formation of a gating scheme from task units to the hidden layer that produces orthogonal representations which are perfectly guarded against interference. Validating the model on previously published human behavioural data revealed that it matches performance of participants who had been trained on blocked or interleaved curricula, and that these performance differences were driven by misestimation of the true category boundary.

Geometrical representation of serial order in working memory

Encoding a sequence relies on one's memory for ordinal succession of events and is critical for episodic memory, spatial navigation, language, and other cognitive functions. Investigating the neural mechanisms underlying sequence working memory in the macaque prefrontal cortex, Xie et al. (Science, 375, 632-639, 2022) uncovered a novel integrated representation of temporal and spatial information in different subspaces of a high-dimensional neural state space, offering broad implications across comparative cognition and neuroscience.

Geometry of spiking patterns in early visual cortex: a topological data analytic approach

In the brain, spiking patterns live in a high-dimensional space of neurons and time. Thus, determining the intrinsic structure of this space presents a theoretical and experimental challenge. To address this challenge, we introduce a new framework for applying topological data analysis (TDA) to spike train data and use it to determine the geometry of spiking patterns in the visual cortex. Key to our approach is a parametrized family of distances based on the timing of spikes that quantifies the dissimilarity between neuronal responses. We applied TDA to visually driven single-unit and multiple single-unit spiking activity in macaque V1 and V2. TDA across timescales reveals a common geometry for spiking patterns in V1 and V2 which, among simple models, is most similar to that of a low-dimensional space endowed with Euclidean or hyperbolic geometry with modest curvature. Remarkably, the inferred geometry depends on timescale and is clearest for the timescales that are important for encoding contrast, orientation and spatial correlations.

A Recurrent Neural Network Model Accounts for Both Timing and Working Memory Components of an Interval Discrimination Task

Interval discrimination is of fundamental importance to many forms of sensory processing, including speech and music. Standard interval discrimination tasks require comparing two intervals separated in time, and thus include both working memory (WM) and timing components. Models of interval discrimination invoke separate circuits for the timing and WM components. Here we examine if, in principle, the same recurrent neural network can implement both. Using human psychophysics, we first explored the role of the WM component by varying the interstimulus delay. Consistent with previous studies, discrimination was significantly worse for a 250 ms delay, compared to 750 and 1500 ms delays, suggesting that the first interval is stably stored in WM for longer delays. We next successfully trained a recurrent neural network (RNN) on the task, demonstrating that the same network can implement both the timing and WM components. Many units in the RNN were tuned to specific intervals during the sensory epoch, and others encoded the first interval during the delay period. Overall, the encoding strategy was consistent with the notion of mixed selectivity. Units generally encoded more interval information during the sensory epoch than in the delay period, reflecting categorical encoding of short versus long in WM, rather than encoding of the specific interval. Our results demonstrate that, in contrast to standard models of interval discrimination that invoke a separate memory module, the same network can, in principle, solve the timing, WM, and comparison components of an interval discrimination task.

Dynamic and stable population coding of attentional instructions coexist in the prefrontal cortex

A large body of recent work suggests that neural representations in prefrontal cortex (PFC) are changing over time to adapt to task demands. However, it remains unclear whether and how such dynamic coding schemes depend on the encoded variable and are influenced by anatomical constraints. Using a cued attention task and multivariate classification methods, we show that neuronal ensembles in PFC encode and retain in working memory spatial and color attentional instructions in an anatomically specific manner. Spatial instructions could be decoded both from the frontal eye field (FEF) and the ventrolateral PFC (vlPFC) population, albeit more robustly from FEF, whereas color instructions were decoded more robustly from vlPFC. Decoding spatial and color information from vlPFC activity in the high-dimensional state space indicated stronger dynamics for color, across the cue presentation and memory periods. The change in the color code was largely due to rapid changes in the network state during the transition to the delay period. However, we found that dynamic vlPFC activity contained time-invariant color information within a low-dimensional subspace of neural activity that allowed for stable decoding of color across time. Furthermore, spatial attention influenced decoding of stimuli features profoundly in vlPFC, but less so in visual area V4. Overall, our results suggest that dynamic population coding of attentional instructions within PFC is shaped by anatomical constraints and can coexist with stable subspace coding that allows time-invariant decoding of information about the future target.

Muscimol-induced inactivation of the ventral prefrontal cortex impairs counting performance in rhesus monkeys

Numbers are one of the three basic concepts of human abstract thinking. When human beings count, they often point to things, one by one, and read numbers in a positive integer column. The prefrontal cortex plays a wide range of roles in executive functions, including active maintenance and achievement of goals, adaptive coding and exertion of general intelligence, and completion of time complexity events. Nonhuman animals do not use number names, such as "one, two, three," or numerals, such as "1, 2, 3" to "count" in the same way as humans do. Our previous study established an animal model of counting in monkeys. Here, we used this model to determine whether the prefrontal cortex participates in counting in monkeys. Two 5-year-old female rhesus monkeys (macaques), weighing 5.0 kg and 5.5 kg, were selected to train in a counting task, counting from 1 to 5. When their counting task performance stabilized, we performed surgery on the prefrontal cortex to implant drug delivery tubes. After allowing the monkeys' physical condition and counting performance to recover, we injected either muscimol or normal saline into their dorsal and ventral prefrontal cortex. Thereafter, we observed their counting task performance and analyzed the error types and reaction time during the counting task. The monkeys' performance in the counting task decreased significantly after muscimol injection into the ventral prefrontal cortex; however, it was not affected after saline injection into the ventral prefrontal cortex, or after muscimol or saline injection into the dorsal prefrontal cortex. The ventral prefrontal cortex of the monkey is necessary for counting performance.

Revisiting the relation between syntax, action, and left BA44

Among the many lines of research that have been exploring how embodiment contributes to cognition, one focuses on how the neural substrates of language may be shared, or at least closely coupled, with those of action. This paper revisits a particular proposal that has received considerable attention—namely, that the forms of hierarchical sequencing that characterize both linguistic syntax and goal-directed action are underpinned partly by common mechanisms in left Brodmann area (BA) 44, a cortical region that is not only classically regarded as part of Broca’s area, but is also a core component of the human Mirror Neuron System. First, a recent multi-participant, multi-round debate about this proposal is summarized together with some other relevant findings. This review reveals that while the proposal is supported by a variety of theoretical arguments and empirical results, it still faces several challenges. Next, a narrower application of the proposal is discussed, specifically involving the basic word order of subject (S), object (O), and verb (V) in simple transitive clauses. Most languages are either SOV or SVO, and, building on prior work, it is argued that these strong syntactic tendencies derive from how left BA44 represents the sequential-hierarchical structure of goal-directed actions. Finally, with the aim of clarifying what it might mean for syntax and action to have “common” neural mechanisms in left BA44, two different versions of the main proposal are distinguished. Hypothesis 1 states that the very same neural mechanisms in left BA44 subserve some aspects of hierarchical sequencing for syntax and action, whereas Hypothesis 2 states that anatomically distinct but functionally parallel neural mechanisms in left BA44 subserve some aspects of hierarchical sequencing for syntax and action. Although these two hypotheses make different predictions, at this point neither one has significantly more explanatory power than the other, and further research is needed to elaborate and test them.

Two-photon calcium imaging of neuronal activity

In vivo two-photon calcium imaging (2PCI) is a technique used for recording neuronal activity in the intact brain. It is based on the principle that, when neurons fire action potentials, intracellular calcium levels rise, which can be detected using fluorescent molecules that bind to calcium. This Primer is designed for scientists who are considering embarking on experiments with 2PCI. We provide the reader with a background on the basic concepts behind calcium imaging and on the reasons why 2PCI is an increasingly powerful and versatile technique in neuroscience. The Primer explains the different steps involved in experiments with 2PCI, provides examples of what ideal preparations should look like and explains how data are analysed. We also discuss some of the current limitations of the technique, and the types of solutions to circumvent them. Finally, we conclude by anticipating what the future of 2PCI might look like, emphasizing some of the analysis pipelines that are being developed and international efforts for data sharing.

Attractive serial dependence overcomes repulsive neuronal adaptation

Sensory responses and behavior are strongly shaped by stimulus history. For example, perceptual reports are sometimes biased toward previously viewed stimuli (serial dependence). While behavioral studies have pointed to both perceptual and postperceptual origins of this phenomenon, neural data that could elucidate where these biases emerge is limited. We recorded functional magnetic resonance imaging (fMRI) responses while human participants (male and female) performed a delayed orientation discrimination task. While behavioral reports were attracted to the previous stimulus, response patterns in visual cortex were repelled. We reconciled these opposing neural and behavioral biases using a model where both sensory encoding and readout are shaped by stimulus history. First, neural adaptation reduces redundancy at encoding and leads to the repulsive biases that we observed in visual cortex. Second, our modeling work suggest that serial dependence is induced by readout mechanisms that account for adaptation in visual cortex. According to this account, the visual system can simultaneously improve efficiency via adaptation while still optimizing behavior based on the temporal structure of natural stimuli.

Symbols and mental programs: a hypothesis about human singularity

Natural language is often seen as the single factor that explains the cognitive singularity of the human species. Instead, we propose that humans possess multiple internal languages of thought, akin to computer languages, which encode and compress structures in various domains (mathematics, music, shape…). These languages rely on cortical circuits distinct from classical language areas. Each is characterized by: (i) the discretization of a domain using a small set of symbols, and (ii) their recursive composition into mental programs that encode nested repetitions with variations. In various tasks of elementary shape or sequence perception, minimum description length in the proposed languages captures human behavior and brain activity, whereas non-human primate data are captured by simpler nonsymbolic models. Our research argues in favor of discrete symbolic models of human thought.

Integrated neural dynamics for behavioural decisions and attentional competition in the prefrontal cortex

In the behaving monkey, complex neural dynamics in the prefrontal cortex contribute to context-dependent decisions and attentional competition. We used demixed principal component analysis to track prefrontal activity dynamics in a cued target detection task. In this task, the animal combined identity of a visual object with a prior instruction cue to determine a target/nontarget decision. From population activity, we extracted principal components for each task feature and examined their time course and sensitivity to stimulus and task variations. For displays containing a single choice object in left or right hemifield, object identity, cue identity and decision were all encoded in population activity, with different dynamics and lateralisation. Object information peaked at 100-200 ms from display onset and was largely confined to the contralateral hemisphere. Cue information was weaker and present even prior to display onset. Integrating information from cue and object, decision information arose more slowly and was bilateral. Individual neurons contributed independently to coding of the three task features. The analysis was then extended to displays with a target in one hemifield and a competing distractor in the other. In this case, the data suggest that each hemisphere initially encoded the identity of the contralateral object. The distractor representation was then rapidly suppressed, with the final target decision again encoded bilaterally. The results show how information is coded along task-related dimensions while competition is resolved and suggest how information flows within and across frontal lobes to implement a learned behavioural decision.

Hierarchical Reinforcement Learning, Sequential Behavior, and the Dorsal Frontostriatal System

To effectively behave within ever-changing environments, biological agents must learn and act at varying hierarchical levels such that a complex task may be broken down into more tractable subtasks. Hierarchical reinforcement learning (HRL) is a computational framework that provides an understanding of this process by combining sequential actions into one temporally extended unit called an option. However, there are still open questions within the HRL framework, including how options are formed and how HRL mechanisms might be realized within the brain. In this review, we propose that the existing human motor sequence literature can aid in understanding both of these questions. We give specific emphasis to visuomotor sequence learning tasks such as the discrete sequence production task and the M × N (M steps × N sets) task to understand how hierarchical learning and behavior manifest across sequential action tasks as well as how the dorsal cortical-subcortical circuitry could support this kind of behavior. This review highlights how motor chunks within a motor sequence can function as HRL options. Furthermore, we aim to merge findings from motor sequence literature with reinforcement learning perspectives to inform experimental design in each respective subfield.