Dynamic divisive normalization predicts time-varying value coding in decision-related circuits

A dynamic scale-mixture model of motion in natural scenes

Some of the most important tasks of visual and motor systems involve estimating the motion of objects and tracking them over time. Such systems evolved to meet the behavioral needs of the organism in its natural environment, and may therefore be adapted to the statistics of motion it is likely to encounter. By tracking the movement of individual points in movies of natural scenes, we begin to identify common properties of natural motion across scenes. As expected, objects in natural scenes move in a persistent fashion, with velocity correlations lasting hundreds of milliseconds. More subtly, but crucially, we find that the observed velocity distributions are heavy-tailed and can be modeled as a Gaussian scale-mixture. Extending this model to the time domain leads to a dynamic scale-mixture model, consisting of a Gaussian process multiplied by a positive scalar quantity with its own independent dynamics. Dynamic scaling of velocity arises naturally as a consequence of changes in object distance from the observer, and may approximate the effects of changes in other parameters governing the motion in a given scene. This modeling and estimation framework has implications for the neurobiology of sensory and motor systems, which need to cope with these fluctuations in scale in order to represent motion efficiently and drive fast and accurate tracking behavior.

Origins of noise in both improving and degrading decision making

Noise is a fundamental problem for information processing in neural systems. In decision-making, noise is assumed to have a primary role in errors and stochastic choice behavior. However, little is known about how noise arising from different sources contributes to value coding and choice behaviors, especially when it interacts with neural computation. Here we examine how noise arising early versus late in the choice process differentially impacts context-dependent choice behavior. We found in model simulations that early and late noise predict opposing context effects: under early noise, contextual information enhances choice accuracy; while under late noise, context degrades choice accuracy. Furthermore, we verified these opposing predictions in experimental human choice behavior. Manipulating early and late noise - by inducing uncertainty in option values and controlling time pressure - produced dissociable positive and negative context effects. These findings reconcile controversial experimental findings in the literature reporting either context-driven impairments or improvements in choice performance, suggesting a unified mechanism for context-dependent choice. More broadly, these findings highlight how different sources of noise can interact with neural computations to differentially modulate behavior.

Significance

The current study addresses the role of noise origin in decision-making, reconciling controversies around how decision-making is impacted by context. We demonstrate that different types of noise - either arising early during evaluation or late during option comparison - leads to distinct results: with early noise, context enhances choice accuracy, while with late noise, context impairs it. Understanding these dynamics offers potential strategies for improving decision-making in noisy environments and refining existing neural computation models. Overall, our findings advance our understanding of how neural systems handle noise in essential cognitive tasks, suggest a beneficial role for contextual modulation under certain conditions, and highlight the profound implications of noise structure in decision-making.

Computational model links normalization to chemoarchitecture in the human visual system

A goal of cognitive neuroscience is to provide computational accounts of brain function. Canonical computations-mathematical operations used by the brain in many contexts-fulfill broad information-processing needs by varying their algorithmic parameters. A key question concerns the identification of biological substrates for these computations and their algorithms. Chemoarchitecture-the spatial distribution of neurotransmitter receptor densities-shapes brain function. Here, we propose that local variations in specific receptor densities implement algorithmic modulations of canonical computations. To test this hypothesis, we combine mathematical modeling of brain responses with chemoarchitecture data. We compare parameters of divisive normalization obtained from 7-tesla functional magnetic resonance imaging with receptor density maps obtained from positron emission tomography. We find evidence that serotonin and γ-aminobutyric acid receptor densities are the biological substrate for algorithmic modulations of divisive normalization in the human visual system. Our model links computational and biological levels of vision, explaining how canonical computations allow the brain to fulfill broad information-processing needs.

A multivariate brain signature for reward

The processing of reinforcers and punishers is crucial to adapt to an ever changing environment and its dysregulation is prevalent in mental health and substance use disorders. While many human brain measures related to reward have been based on activity in individual brain regions, recent studies indicate that many affective and motivational processes are encoded in distributed systems that span multiple regions. Consequently, decoding these processes using individual regions yields small effect sizes and limited reliability, whereas predictive models based on distributed patterns yield larger effect sizes and excellent reliability. To create such a predictive model for the processes of rewards and losses, termed the Brain Reward Signature (BRS), we trained a model to predict the signed magnitude of monetary rewards on the Monetary Incentive Delay task (MID; N = 39) and achieved a highly significant decoding performance (92% for decoding rewards versus losses). We subsequently demonstrate the generalizability of our signature on another version of the MID in a different sample (92% decoding accuracy; N = 12) and on a gambling task from a large sample (73% decoding accuracy, N = 1084). We further provided preliminary data to characterize the specificity of the signature by illustrating that the signature map generates estimates that significantly differ between rewarding and negative feedback (92% decoding accuracy) but do not differ for conditions that differ in disgust rather than reward in a novel Disgust-Delay Task (N = 39). Finally, we show that passively viewing positive and negatively valenced facial expressions loads positively on our signature, in line with previous studies on morbid curiosity. We thus created a BRS that can accurately predict brain responses to rewards and losses in active decision making tasks, and that possibly relates to information seeking in passive observational tasks.

Intraparietal stimulation disrupts negative distractor effects in human multi-alternative decision-making

There has been debate about whether addition of an irrelevant distractor option to an otherwise binary decision influences which of the two choices is taken. We show that disparate views on this question are reconciled if distractors exert two opposing but not mutually exclusive effects. Each effect predominates in a different part of decision space: (1) a positive distractor effect predicts high-value distractors improve decision-making; (2) a negative distractor effect, of the type associated with divisive normalisation models, entails decreased accuracy with increased distractor values. Here, we demonstrate both distractor effects coexist in human decision making but in different parts of a decision space defined by the choice values. We show disruption of the medial intraparietal area (MIP) by transcranial magnetic stimulation (TMS) increases positive distractor effects at the expense of negative distractor effects. Furthermore, individuals with larger MIP volumes are also less susceptible to the disruption induced by TMS. These findings also demonstrate a causal link between MIP and the impact of distractors on decision-making via divisive normalisation.

Neural Mechanisms That Make Perceptual Decisions Flexible

Neural mechanisms of perceptual decision making have been extensively studied in experimental settings that mimic stable environments with repeating stimuli, fixed rules, and payoffs. In contrast, we live in an ever-changing environment and have varying goals and behavioral demands. To accommodate variability, our brain flexibly adjusts decision-making processes depending on context. Here, we review a growing body of research that explores the neural mechanisms underlying this flexibility. We highlight diverse forms of context dependency in decision making implemented through a variety of neural computations. Context-dependent neural activity is observed in a distributed network of brain structures, including posterior parietal, sensory, motor, and subcortical regions, as well as the prefrontal areas classically implicated in cognitive control. We propose that investigating the distributed network underlying flexible decisions is key to advancing our understanding and discuss a path forward for experimental and theoretical investigations.

Neuronal Response to Reward and Luminance in Macaque LIP During Saccadic Choice

Recent work in decision neuroscience suggests that visual saliency can interact with reward-based choice, and the lateral intraparietal cortex (LIP) is implicated in this process. In this study, we recorded from LIP neurons while monkeys performed a two alternative choice task in which the reward and luminance associated with each offer were varied independently. We discovered that the animal's choice was dictated by the reward amount while the luminance had a marginal effect. In the LIP, neuronal activity corresponded well with the animal's choice pattern, in that a majority of reward-modulated neurons encoded the reward amount in the neuron's preferred hemifield with a positive slope. In contrast, compared to their responses to low luminance, an approximately equal proportion of luminance-sensitive neurons responded to high luminance with increased or decreased activity, leading to a much weaker population-level response. Meanwhile, in the non-preferred hemifield, the strength of encoding for reward amount and luminance was positively correlated, suggesting the integration of these two factors in the LIP. Moreover, neurons encoding reward and luminance were homogeneously distributed along the anterior-posterior axis of the LIP. Overall, our study provides further evidence supporting the neural instantiation of a priority map in the LIP in reward-based decisions.

Asymmetric and adaptive reward coding via normalized reinforcement learning

Learning is widely modeled in psychology, neuroscience, and computer science by prediction error-guided reinforcement learning (RL) algorithms. While standard RL assumes linear reward functions, reward-related neural activity is a saturating, nonlinear function of reward; however, the computational and behavioral implications of nonlinear RL are unknown. Here, we show that nonlinear RL incorporating the canonical divisive normalization computation introduces an intrinsic and tunable asymmetry in prediction error coding. At the behavioral level, this asymmetry explains empirical variability in risk preferences typically attributed to asymmetric learning rates. At the neural level, diversity in asymmetries provides a computational mechanism for recently proposed theories of distributional RL, allowing the brain to learn the full probability distribution of future rewards. This behavioral and computational flexibility argues for an incorporation of biologically valid value functions in computational models of learning and decision-making.

Reinforcement learning models are widely used to characterize reward-driven learning in biological and computational agents. Standard reinforcement learning models use linear value functions, despite strong empirical evidence that biological value representations are nonlinear functions of external rewards. Here, we examine the properties of a biologically-based nonlinear reinforcement learning algorithm employing the canonical divisive normalization function, a neural computation commonly found in sensory, cognitive, and reward coding. We show that this normalized reinforcement learning algorithm implements a simple but powerful control of how reward learning reflects relative gains and losses. This property explains diverse behavioral and neural phenomena, and suggests the importance of using biologically valid value functions in computational models of learning and decision-making.

Efficiently irrational: deciphering the riddle of human choice

For the past half-century, cognitive and social scientists have struggled with the irrationalities of human choice behavior; people consistently make choices that are logically inconsistent. Is human choice behavior evolutionarily adaptive or is it an inefficient patchwork of competing mechanisms? In this review, I present an interdisciplinary synthesis arguing for a novel interpretation: choice is efficiently irrational. Connecting findings across disciplines suggests that observed choice behavior reflects a precise optimization of the trade-off between the costs of increasing the precision of the choice mechanism and the declining benefits that come as precision increases. Under these constraints, a rationally imprecise strategy emerges that works toward optimal efficiency rather than toward optimal rationality. This approach rationalizes many of the puzzling inconsistencies of human choice behavior, explaining why these inconsistencies arise as an optimizing solution in biological choosers.

Normalization by valence and motivational intensity in the sensorimotor cortices (PMd, M1, and S1)

Our brain's ability to represent vast amounts of information, such as continuous ranges of reward spanning orders of magnitude, with limited dynamic range neurons, may be possible due to normalization. Recently our group and others have shown that the sensorimotor cortices are sensitive to reward value. Here we ask if psychological affect causes normalization of the sensorimotor cortices by modulating valence and motivational intensity. We had two non-human primates (NHP) subjects (one male bonnet macaque and one female rhesus macaque) make visually cued grip-force movements while simultaneously cueing the level of possible reward if successful, or timeout punishment, if unsuccessful. We recorded simultaneously from 96 electrodes in each the following: caudal somatosensory, rostral motor, and dorsal premotor cortices (cS1, rM1, PMd). We utilized several normalization models for valence and motivational intensity in all three regions. We found three types of divisive normalized relationships between neural activity and the representation of valence and motivation, linear, sigmodal, and hyperbolic. The hyperbolic relationships resemble receptive fields in psychological affect space, where a unit is susceptible to a small range of the valence/motivational space. We found that these cortical regions have both strong valence and motivational intensity representations.

A dynamic normalization model of temporal attention

Vision is dynamic, handling a continuously changing stream of input, yet most models of visual attention are static. Here, we develop a dynamic normalization model of visual temporal attention and constrain it with new psychophysical human data. We manipulated temporal attention-the prioritization of visual information at specific points in time-to a sequence of two stimuli separated by a variable time interval. Voluntary temporal attention improved perceptual sensitivity only over a specific interval range. To explain these data, we modelled voluntary and involuntary attentional gain dynamics. Voluntary gain enhancement took the form of a limited resource over short time intervals, which recovered over time. Taken together, our theoretical and experimental results formalize and generalize the idea of limited attentional resources across space at a single moment to limited resources across time at a single location.

Divisive normalization unifies disparate response signatures throughout the human visual hierarchy

A canonical neural computation is a mathematical operation applied by the brain in a wide variety of contexts and capable of explaining and unifying seemingly unrelated neural and perceptual phenomena. Here, we use a combination of state-of-the-art experiments (ultra-high-field functional MRI) and mathematical methods (population receptive field [pRF] modeling) to uniquely demonstrate the role of divisive normalization (DN) as the canonical neural computation underlying visuospatial responses throughout the human visual hierarchy. The DN pRF model provides a tool to investigate and interpret the computational processes underlying neural responses in human and animal recordings, but also in clinical and cognitive dimensions.

Neural processing is hypothesized to apply the same mathematical operations in a variety of contexts, implementing so-called canonical neural computations. Divisive normalization (DN) is considered a prime candidate for a canonical computation. Here, we propose a population receptive field (pRF) model based on DN and evaluate it using ultra-high-field functional MRI (fMRI). The DN model parsimoniously captures seemingly disparate response signatures with a single computation, superseding existing pRF models in both performance and biological plausibility. We observe systematic variations in specific DN model parameters across the visual hierarchy and show how they relate to differences in response modulation and visuospatial information integration. The DN model delivers a unifying framework for visuospatial responses throughout the human visual hierarchy and provides insights into its underlying information-encoding computations. These findings extend the role of DN as a canonical computation to neuronal populations throughout the human visual hierarchy.

Dynamic divisive normalization circuits explain and predict change detection in monkey area MT

Sudden changes in visual scenes often indicate important events for behavior. For their quick and reliable detection, the brain must be capable to process these changes as independently as possible from its current activation state. In motion-selective area MT, neurons respond to instantaneous speed changes with pronounced transients, often far exceeding the expected response as derived from their speed tuning profile. We here show that this complex, non-linear behavior emerges from the combined temporal dynamics of excitation and divisive inhibition, and provide a comprehensive mathematical analysis. A central prediction derived from this investigation is that attention increases the steepness of the transient response irrespective of the activation state prior to a stimulus change, and irrespective of the sign of the change (i.e. irrespective of whether the stimulus is accelerating or decelerating). Extracellular recordings of attention-dependent representation of both speed increments and decrements confirmed this prediction and suggest that improved change detection derives from basic computations in a canonical cortical circuitry.

Excitatory-inhibitory tone shapes decision strategies in a hierarchical neural network model of multi-attribute choice

We are constantly faced with decisions between alternatives defined by multiple attributes, necessitating an evaluation and integration of different information sources. Time-varying signals in multiple brain areas are implicated in decision-making; but we lack a rigorous biophysical description of how basic circuit properties, such as excitatory-inhibitory (E/I) tone and cascading nonlinearities, shape attribute processing and choice behavior. Furthermore, how such properties govern choice performance under varying levels of environmental uncertainty is unknown. We investigated two-attribute, two-alternative decision-making in a dynamical, cascading nonlinear neural network with three layers: an input layer encoding choice alternative attribute values; an intermediate layer of modules processing separate attributes; and a final layer producing the decision. Depending on intermediate layer E/I tone, the network displays distinct regimes characterized by linear (I), convex (II) or concave (III) choice indifference curves. In regimes I and II, each option's attribute information is additively integrated. In regime III, time-varying nonlinear operations amplify the separation between offer distributions by selectively attending to the attribute with the larger differences in input values. At low environmental uncertainty, a linear combination most consistently selects higher valued alternatives. However, at high environmental uncertainty, regime III is more likely than a linear operation to select alternatives with higher value. Furthermore, there are conditions where readout from the intermediate layer could be experimentally indistinguishable from the final layer. Finally, these principles are used to examine multi-attribute decisions in systems with reduced inhibitory tone, leading to predictions of different choice patterns and overall performance between those with restrictions on inhibitory tone and neurotypicals.

A recurrent circuit implements normalization, simulating the dynamics of V1 activity

The normalization model has been applied to explain neural activity in diverse neural systems including primary visual cortex (V1). The model's defining characteristic is that the response of each neuron is divided by a factor that includes a weighted sum of activity of a pool of neurons. Despite the success of the normalization model, there are three unresolved issues. 1) Experimental evidence supports the hypothesis that normalization in V1 operates via recurrent amplification, i.e., amplifying weak inputs more than strong inputs. It is unknown how normalization arises from recurrent amplification. 2) Experiments have demonstrated that normalization is weighted such that each weight specifies how one neuron contributes to another's normalization pool. It is unknown how weighted normalization arises from a recurrent circuit. 3) Neural activity in V1 exhibits complex dynamics, including gamma oscillations, linked to normalization. It is unknown how these dynamics emerge from normalization. Here, a family of recurrent circuit models is reported, each of which comprises coupled neural integrators to implement normalization via recurrent amplification with arbitrary normalization weights, some of which can recapitulate key experimental observations of the dynamics of neural activity in V1.

Consistent patterns of distractor effects during decision making

The value of a third potential option or distractor can alter the way in which decisions are made between two other options. Two hypotheses have received empirical support: that a high value distractor improves the accuracy with which decisions between two other options are made and that it impairs accuracy. Recently, however, it has been argued that neither observation is replicable. Inspired by neuroimaging data showing that high value distractors have different impacts on prefrontal and parietal regions, we designed a dual route decision-making model that mimics the neural signals of these regions. Here we show in the dual route model and empirical data that both enhancement and impairment effects are robust phenomena but predominate in different parts of the decision space defined by the options' and the distractor's values. However, beyond these constraints, both effects co-exist under similar conditions. Moreover, both effects are robust and observable in six experiments.

A divisive model of evidence accumulation explains uneven weighting of evidence over time

Divisive normalization has long been used to account for computations in various neural processes and behaviours. The model proposes that inputs into a neural system are divisively normalized by the system’s total activity. More recently, dynamical versions of divisive normalization have been shown to account for how neural activity evolves over time in value-based decision making. Despite its ubiquity, divisive normalization has not been studied in decisions that require evidence to be integrated over time. Such decisions are important when the information is not all available at once. A key feature of such decisions is how evidence is weighted over time, known as the integration kernel. Here, we provide a formal expression for the integration kernel in divisive normalization, and show that divisive normalization quantitatively accounts for 133 human participants’ perceptual decision making behaviour, performing as well as the state-of-the-art Drift Diffusion Model, the predominant model for perceptual evidence accumulation.

Divisive normalization is thought to be a ubiquitous computation in the brain, but has not been studied in decisions that require integrating evidence over time. Here, the authors show in humans that dynamic divisive normalization accounts for the uneven weighting of perceptual evidence over time.

Oscillatory recurrent gated neural integrator circuits (ORGaNICs), a unifying theoretical framework for neural dynamics

Working memory is an example of a cognitive and neural process that is not static but evolves dynamically with changing sensory inputs; another example is motor preparation and execution. We introduce a theoretical framework for neural dynamics, based on oscillatory recurrent gated neural integrator circuits (ORGaNICs), and apply it to simulate key phenomena of working memory and motor control. The model circuits simulate neural activity with complex dynamics, including sequential activity and traveling waves of activity, that manipulate (as well as maintain) information during working memory. The same circuits convert spatial patterns of premotor activity to temporal profiles of motor control activity and manipulate (e.g., time warp) the dynamics. Derivative-like recurrent connectivity, in particular, serves to manipulate and update internal models, an essential feature of working memory and motor execution. In addition, these circuits incorporate recurrent normalization, to ensure stability over time and robustness with respect to perturbations of synaptic weights.

Predicting neuronal dynamics with a delayed gain control model

Visual neurons respond to static images with specific dynamics: neuronal responses sum sub-additively over time, reduce in amplitude with repeated or sustained stimuli (neuronal adaptation), and are slower at low stimulus contrast. Here, we propose a simple model that predicts these seemingly disparate response patterns observed in a diverse set of measurements-intracranial electrodes in patients, fMRI, and macaque single unit spiking. The model takes a time-varying contrast time course of a stimulus as input, and produces predicted neuronal dynamics as output. Model computation consists of linear filtering, expansive exponentiation, and a divisive gain control. The gain control signal relates to but is slower than the linear signal, and this delay is critical in giving rise to predictions matched to the observed dynamics. Our model is simpler than previously proposed related models, and fitting the model to intracranial EEG data uncovers two regularities across human visual field maps: estimated linear filters (temporal receptive fields) systematically differ across and within visual field maps, and later areas exhibit more rapid and substantial gain control. The model is further generalizable to account for dynamics of contrast-dependent spike rates in macaque V1, and amplitudes of fMRI BOLD in human V1.

Economic Decisions through Circuit Inhibition

Economic choices between goods are thought to rely on the orbitofrontal cortex (OFC), but the decision mechanisms remain poorly understood. To shed light on this fundamental issue, we recorded from the OFC of monkeys choosing between two juices offered sequentially. An analysis of firing rates across time windows revealed the presence of different groups of neurons similar to those previously identified under simultaneous offers. This observation suggested that economic decisions in the two modalities are formed in the same neural circuit. We then examined several hypotheses on the decision mechanisms. OFC neurons encoded good identities and values in a juice-based representation (labeled lines). Contrary to previous assessments, our data argued against the idea that decisions rely on mutual inhibition at the level of offer values. In fact, we showed that previous arguments for mutual inhibition were confounded by differences in value ranges. Instead, decisions seemed to involve mechanisms of circuit inhibition, whereby each offer value indirectly inhibited neurons encoding the opposite choice outcome. Our results reconcile a variety of previous findings and provide a general account for the neuronal underpinnings of economic choices.

Why context matters? Divisive normalization and canonical microcircuits in psychiatric disorders

Neural activity on cellular, regional, and behavioral levels shows context-dependence. Here we suggest the processing of input-output relationships in terms divisive normalization (DN), including (i) summing/averaging inputs and (ii) normalizing output against input stages, as a computational mechanism to underlie context-dependence. Input summation and output normalization are mediated by input-output relationships in canonical microcircuits (CM). DN/CM are altered in psychiatric disorders like schizophrenia or depression whose various symptoms can be characterized by abnormal context-dependence.

Categorical encoding of decision variables in orbitofrontal cortex

A fundamental and recurrent question in systems neuroscience is that of assessing what variables are encoded by a given population of neurons. Such assessments are often challenging because neurons in one brain area may encode multiple variables, and because neuronal representations might be categorical or non-categorical. These issues are particularly pertinent to the representation of decision variables in the orbitofrontal cortex (OFC)-an area implicated in economic choices. Here we present a new algorithm to assess whether a neuronal representation is categorical or non-categorical, and to identify the encoded variables if the representation is indeed categorical. The algorithm is based on two clustering procedures, one variable-independent and the other variable-based. The two partitions are then compared through adjusted mutual information. The present algorithm overcomes limitations of previous approaches and is widely applicable. We tested the algorithm on synthetic data and then used it to examine neuronal data recorded in the primate OFC during economic decisions. Confirming previous assessments, we found the neuronal representation in OFC to be categorical in nature. We also found that neurons in this area encode the value of individual offers, the binary choice outcome and the chosen value. In other words, during economic choice, neurons in the primate OFC encode decision variables in a categorical way.

Relating Divisive Normalization to Neuronal Response Variability

Cortical responses to repeated presentations of a sensory stimulus are variable. This variability is sensitive to several stimulus dimensions, suggesting that it may carry useful information beyond the average firing rate. Many experimental manipulations that affect response variability are also known to engage divisive normalization, a widespread operation that describes neuronal activity as the ratio of a numerator (representing the excitatory stimulus drive) and denominator (the normalization signal). Although it has been suggested that normalization affects response variability, we lack a quantitative framework to determine the relation between the two. Here we extend the standard normalization model, by treating the numerator and the normalization signal as variable quantities. The resulting model predicts a general stabilizing effect of normalization on neuronal responses, and allows us to infer the single-trial normalization strength, a quantity that cannot be measured directly. We test the model on neuronal responses to stimuli of varying contrast, recorded in primary visual cortex of male macaques. We find that neurons that are more strongly normalized fire more reliably, and response variability and pairwise noise correlations are reduced during trials in which normalization is inferred to be strong. Our results thus suggest a novel functional role for normalization, namely, modulating response variability. Our framework could enable a direct quantification of the impact of single-trial normalization strength on the accuracy of perceptual judgments, and can be readily applied to other sensory and nonsensory factors.SIGNIFICANCE STATEMENT Divisive normalization is a widespread neural operation across sensory and nonsensory brain areas, which describes neuronal responses as the ratio between the excitatory drive to the neuron and a normalization signal. Normalization plays a key role in several important computations, including adjusting the neuron's dynamic range, reducing redundancy, and facilitating probabilistic inference. However, the relation between normalization and neuronal response variability (a fundamental aspect of neural coding) remains unclear. Here we develop a new model and test it on primary visual cortex responses. We show that normalization has a stabilizing effect on neuronal activity, beyond the known suppression of firing rate. This modulation of variability suggests a new functional role for normalization in neural coding and perception.

Reference effects on decision-making elicited by previous rewards

Substantial evidence has highlighted reference effects occurring during decision-making, whereby subjective value is not calculated in absolute terms but relative to the distribution of rewards characterizing a context. Among these, within-choice effects are exerted by options simultaneously available during choice. These should be distinguished from between-choice effects, which depend on the distribution of options presented in the past. Influential theories on between-choice effects include Decision-by-Sampling, Expectation-as-Reference and Divisive Normalization. Surprisingly, previous literature has focused on each theory individually disregarding the others. Thus, similarities and differences among theories remain to be systematically examined. Here we fill this gap by offering an overview of the state-of-the-art of research about between-choice reference effects. Our comparison of alternative theories shows that, at present, none of them is able to account for the full range of empirical data. To address this, we propose a model inspired by previous perspectives and based on a logistic framework, hence called logistic model of subjective value. Predictions of the model are analysed in detail about reference effects and risky decision-making. We conclude that our proposal offers a compelling framework for interpreting the multifaceted manifestations of between-choice reference effects.

Optimal policy for multi-alternative decisions

Everyday decisions frequently require choosing among multiple alternatives. Yet the optimal policy for such decisions is unknown. Here we derive the normative policy for general multi-alternative decisions. This strategy requires evidence accumulation to nonlinear, time-dependent bounds that trigger choices. A geometric symmetry in those boundaries allows the optimal strategy to be implemented by a simple neural circuit involving normalization with fixed decision bounds and an urgency signal. The model captures several key features of the response of decision-making neurons as well as the increase in reaction time as a function of the number of alternatives, known as Hick's law. In addition, we show that in the presence of divisive normalization and internal variability, our model can account for several so-called 'irrational' behaviors, such as the similarity effect as well as the violation of both the independence of irrelevant alternatives principle and the regularity principle.

Reference-point centering and range-adaptation enhance human reinforcement learning at the cost of irrational preferences

In economics and perceptual decision-making contextual effects are well documented, where decision weights are adjusted as a function of the distribution of stimuli. Yet, in reinforcement learning literature whether and how contextual information pertaining to decision states is integrated in learning algorithms has received comparably little attention. Here, we investigate reinforcement learning behavior and its computational substrates in a task where we orthogonally manipulate outcome valence and magnitude, resulting in systematic variations in state-values. Model comparison indicates that subjects' behavior is best accounted for by an algorithm which includes both reference point-dependence and range-adaptation-two crucial features of state-dependent valuation. In addition, we find that state-dependent outcome valuation progressively emerges, is favored by increasing outcome information and correlated with explicit understanding of the task structure. Finally, our data clearly show that, while being locally adaptive (for instance in negative valence and small magnitude contexts), state-dependent valuation comes at the cost of seemingly irrational choices, when options are extrapolated out from their original contexts.

Multiple timescales of normalized value coding underlie adaptive choice behavior

Adaptation is a fundamental process crucial for the efficient coding of sensory information. Recent evidence suggests that similar coding principles operate in decision-related brain areas, where neural value coding adapts to recent reward history. However, the circuit mechanism for value adaptation is unknown, and the link between changes in adaptive value coding and choice behavior is unclear. Here we show that choice behavior in nonhuman primates varies with the statistics of recent rewards. Consistent with efficient coding theory, decision-making shows increased choice sensitivity in lower variance reward environments. Both the average adaptation effect and across-session variability are explained by a novel multiple timescale dynamical model of value representation implementing divisive normalization. The model predicts empirical variance-driven changes in behavior despite having no explicit knowledge of environmental statistics, suggesting that distributional characteristics can be captured by dynamic model architectures. These findings highlight the importance of treating decision-making as a dynamic process and the role of normalization as a unifying computation for contextual phenomena in choice.

Continuous aesthetic judgment of image sequences

Perceptual judgments are said to be reference-dependent as they change on the basis of recent experiences. Here we quantify sequence effects within two types of aesthetic judgments: (i) individual ratings of single images (during self-paced trials) and (ii) continuous ratings of image sequences. As in the case of known contrast effects, trial-by-trial aesthetic responses are negatively correlated with judgments made toward the preceding image. During continuous judgment, a different type of bias is observed. The onset of change within a sequence introduces a persistent increase in ratings (relative to when the same images are judged in isolation). Furthermore, subjects indicate adjustment patterns and choices that selectively favor sequences that are rich in change. Sequence effects in aesthetic judgments thus differ greatly depending on the continuity and arrangement of presented stimuli. The effects highlighted here are important in understanding sustained aesthetic responses over time, such as those elicited during choreographic and musical arrangements. In contrast, standard measurements of aesthetic responses (over trials) may represent a series of distinct aesthetic experiences (e.g., viewing artworks in a museum).

A transformation from temporal to ensemble coding in a model of piriform cortex

Different coding strategies are used to represent odor information at various stages of the mammalian olfactory system. A temporal latency code represents odor identity in olfactory bulb (OB), but this temporal information is discarded in piriform cortex (PCx) where odor identity is instead encoded through ensemble membership. We developed a spiking PCx network model to understand how this transformation is implemented. In the model, the impact of OB inputs activated earliest after inhalation is amplified within PCx by diffuse recurrent collateral excitation, which then recruits strong, sustained feedback inhibition that suppresses the impact of later-responding glomeruli. We model increasing odor concentrations by decreasing glomerulus onset latencies while preserving their activation sequences. This produces a multiplexed cortical odor code in which activated ensembles are robust to concentration changes while concentration information is encoded through population synchrony. Our model demonstrates how PCx circuitry can implement multiplexed ensemble-identity/temporal-concentration odor coding.

The Role of the Lateral Intraparietal Area in (the Study of) Decision Making

Over the past two decades, neurophysiological responses in the lateral intraparietal area (LIP) have received extensive study for insight into decision making. In a parallel manner, inferred cognitive processes have enriched interpretations of LIP activity. Because of this bidirectional interplay between physiology and cognition, LIP has served as fertile ground for developing quantitative models that link neural activity with decision making. These models stand as some of the most important frameworks for linking brain and mind, and they are now mature enough to be evaluated in finer detail and integrated with other lines of investigation of LIP function. Here, we focus on the relationship between LIP responses and known sensory and motor events in perceptual decision-making tasks, as assessed by correlative and causal methods. The resulting sensorimotor-focused approach offers an account of LIP activity as a multiplexed amalgam of sensory, cognitive, and motor-related activity, with a complex and often indirect relationship to decision processes. Our data-driven focus on multiplexing (and de-multiplexing) of various response components can complement decision-focused models and provides more detailed insight into how neural signals might relate to cognitive processes such as decision making.

Inverted activity patterns in ventromedial prefrontal cortex during value-guided decision-making in a less-is-more task

Ventromedial prefrontal cortex has been linked to choice evaluation and decision-making in humans but understanding the role it plays is complicated by the fact that little is known about the corresponding area of the macaque brain. We recorded activity in macaques using functional magnetic resonance imaging during two very different value-guided decision-making tasks. In both cases ventromedial prefrontal cortex activity reflected subjective choice values during decision-making just as in humans but the relationship between the blood oxygen level-dependent signal and both decision-making and choice value was inverted and opposite to the relationship seen in humans. In order to test whether the ventromedial prefrontal cortex activity related to choice values is important for decision-making we conducted an additional lesion experiment; lesions that included the same ventromedial prefrontal cortex region disrupted normal subjective evaluation of choices during decision-making.

Ventromedial prefrontal cortex in humans shows functional magnetic resonance imaging signals related to the subjective values of choices that are taken during decision-making as well as task-negative signals. Here, the authors report that in macaque ventromedial prefrontal cortex both activity patterns are inverted and lesions of this area disrupt subjective choice evaluation.

Normalized value coding explains dynamic adaptation in the human valuation process

The notion of subjective value is central to choice theories in ecology, economics, and psychology, serving as an integrated decision variable by which options are compared. Subjective value is often assumed to be an absolute quantity, determined in a static manner by the properties of an individual option. Recent neurobiological studies, however, have shown that neural value coding dynamically adapts to the statistics of the recent reward environment, introducing an intrinsic temporal context dependence into the neural representation of value. Whether valuation exhibits this kind of dynamic adaptation at the behavioral level is unknown. Here, we show that the valuation process in human subjects adapts to the history of previous values, with current valuations varying inversely with the average value of recently observed items. The dynamics of this adaptive valuation are captured by divisive normalization, linking these temporal context effects to spatial context effects in decision making as well as spatial and temporal context effects in perception. These findings suggest that adaptation is a universal feature of neural information processing and offer a unifying explanation for contextual phenomena in fields ranging from visual psychophysics to economic choice.

Optimal coding and neuronal adaptation in economic decisions

During economic decisions, offer value cells in orbitofrontal cortex (OFC) encode the values of offered goods. Furthermore, their tuning functions adapt to the range of values available in any given context. A fundamental and open question is whether range adaptation is behaviorally advantageous. Here we present a theory of optimal coding for economic decisions. We propose that the representation of offer values is optimal if it ensures maximal expected payoff. In this framework, we examine offer value cells in non-human primates. We show that their responses are quasi-linear even when optimal tuning functions are highly non-linear. Most importantly, we demonstrate that for linear tuning functions range adaptation maximizes the expected payoff. Thus value coding in OFC is functionally rigid (linear tuning) but parametrically plastic (range adaptation with optimal gain). Importantly, the benefit of range adaptation outweighs the cost of functional rigidity. While generally suboptimal, linear tuning may facilitate transitive choices.

Value-based decision making via sequential sampling with hierarchical competition and attentional modulation

In principle, formal dynamical models of decision making hold the potential to represent fundamental computations underpinning value-based (i.e., preferential) decisions in addition to perceptual decisions. Sequential-sampling models such as the race model and the drift-diffusion model that are grounded in simplicity, analytical tractability, and optimality remain popular, but some of their more recent counterparts have instead been designed with an aim for more feasibility as architectures to be implemented by actual neural systems. Connectionist models are proposed herein at an intermediate level of analysis that bridges mental phenomena and underlying neurophysiological mechanisms. Several such models drawing elements from the established race, drift-diffusion, feedforward-inhibition, divisive-normalization, and competing-accumulator models were tested with respect to fitting empirical data from human participants making choices between foods on the basis of hedonic value rather than a traditional perceptual attribute. Even when considering performance at emulating behavior alone, more neurally plausible models were set apart from more normative race or drift-diffusion models both quantitatively and qualitatively despite remaining parsimonious. To best capture the paradigm, a novel six-parameter computational model was formulated with features including hierarchical levels of competition via mutual inhibition as well as a static approximation of attentional modulation, which promotes "winner-take-all" processing. Moreover, a meta-analysis encompassing several related experiments validated the robustness of model-predicted trends in humans' value-based choices and concomitant reaction times. These findings have yet further implications for analysis of neurophysiological data in accordance with computational modeling, which is also discussed in this new light.

A unifying Bayesian account of contextual effects in value-based choice

Empirical evidence suggests the incentive value of an option is affected by other options available during choice and by options presented in the past. These contextual effects are hard to reconcile with classical theories and have inspired accounts where contextual influences play a crucial role. However, each account only addresses one or the other of the empirical findings and a unifying perspective has been elusive. Here, we offer a unifying theory of context effects on incentive value attribution and choice based on normative Bayesian principles. This formulation assumes that incentive value corresponds to a precision-weighted prediction error, where predictions are based upon expectations about reward. We show that this scheme explains a wide range of contextual effects, such as those elicited by other options available during choice (or within-choice context effects). These include both conditions in which choice requires an integration of multiple attributes and conditions where a multi-attribute integration is not necessary. Moreover, the same scheme explains context effects elicited by options presented in the past or between-choice context effects. Our formulation encompasses a wide range of contextual influences (comprising both within- and between-choice effects) by calling on Bayesian principles, without invoking ad-hoc assumptions. This helps clarify the contextual nature of incentive value and choice behaviour and may offer insights into psychopathologies characterized by dysfunctional decision-making, such as addiction and pathological gambling.

Learning Contextual Reward Expectations for Value Adaptation

Substantial evidence indicates that subjective value is adapted to the statistics of reward expected within a given temporal context. However, how these contextual expectations are learned is poorly understood. To examine such learning, we exploited a recent observation that participants performing a gambling task adjust their preferences as a function of context. We show that, in the absence of contextual cues providing reward information, an average reward expectation was learned from recent past experience. Learning dependent on contextual cues emerged when two contexts alternated at a fast rate, whereas both cue-independent and cue-dependent forms of learning were apparent when two contexts alternated at a slower rate. Motivated by these behavioral findings, we reanalyzed a previous fMRI data set to probe the neural substrates of learning contextual reward expectations. We observed a form of reward prediction error related to average reward such that, at option presentation, activity in ventral tegmental area/substantia nigra and ventral striatum correlated positively and negatively, respectively, with the actual and predicted value of options. Moreover, an inverse correlation between activity in ventral tegmental area/substantia nigra (but not striatum) and predicted option value was greater in participants showing enhanced choice adaptation to context. The findings help understanding the mechanisms underlying learning of contextual reward expectation.

Pattern Adaptation and Normalization Reweighting

Adaptation to an oriented stimulus changes both the gain and preferred orientation of neural responses in V1. Neurons tuned near the adapted orientation are suppressed, and their preferred orientations shift away from the adapter. We propose a model in which weights of divisive normalization are dynamically adjusted to homeostatically maintain response products between pairs of neurons. We demonstrate that this adjustment can be performed by a very simple learning rule. Simulations of this model closely match existing data from visual adaptation experiments. We consider several alternative models, including variants based on homeostatic maintenance of response correlations or covariance, as well as feedforward gain-control models with multiple layers, and we demonstrate that homeostatic maintenance of response products provides the best account of the physiological data.

SIGNIFICANCE STATEMENT

Adaptation is a phenomenon throughout the nervous system in which neural tuning properties change in response to changes in environmental statistics. We developed a model of adaptation that combines normalization (in which a neuron's gain is reduced by the summed responses of its neighbors) and Hebbian learning (in which synaptic strength, in this case divisive normalization, is increased by correlated firing). The model is shown to account for several properties of adaptation in primary visual cortex in response to changes in the statistics of contour orientation.

Adaptive Value Normalization in the Prefrontal Cortex Is Reduced by Memory Load

Adaptation facilitates neural representation of a wide range of diverse inputs, including reward values. Adaptive value coding typically relies on contextual information either obtained from the environment or retrieved from and maintained in memory. However, it is unknown whether having to retrieve and maintain context information modulates the brain's capacity for value adaptation. To address this issue, we measured hemodynamic responses of the prefrontal cortex (PFC) in two studies on risky decision-making. In each trial, healthy human subjects chose between a risky and a safe alternative; half of the participants had to remember the risky alternatives, whereas for the other half they were presented visually. The value of safe alternatives varied across trials. PFC responses adapted to contextual risk information, with steeper coding of safe alternative value in lower-risk contexts. Importantly, this adaptation depended on working memory load, such that response functions relating PFC activity to safe values were steeper with presented versus remembered risk. An independent second study replicated the findings of the first study and showed that similar slope reductions also arose when memory maintenance demands were increased with a secondary working memory task. Formal model comparison showed that a divisive normalization model fitted effects of both risk context and working memory demands on PFC activity better than alternative models of value adaptation, and revealed that reduced suppression of background activity was the critical parameter impairing normalization with increased memory maintenance demand. Our findings suggest that mnemonic processes can constrain normalization of neural value representations.

Computational Architecture of the Parieto-Frontal Network Underlying Cognitive-Motor Control in Monkeys

The statistical structure of intrinsic parietal and parieto-frontal connectivity in monkeys was studied through hierarchical cluster analysis. Based on their inputs, parietal and frontal areas were grouped into different clusters, including a variable number of areas that in most instances occupied contiguous architectonic fields. Connectivity tended to be stronger locally: that is, within areas of the same cluster. Distant frontal and parietal areas were targeted through connections that in most instances were reciprocal and often of different strength. These connections linked parietal and frontal clusters formed by areas sharing basic functional properties. This led to five different medio-laterally oriented pillar domains spanning the entire extent of the parieto-frontal system, in the posterior parietal, anterior parietal, cingulate, frontal, and prefrontal cortex. Different information processing streams could be identified thanks to inter-domain connectivity. These streams encode fast hand reaching and its control, complex visuomotor action spaces, hand grasping, action/intention recognition, oculomotor intention and visual attention, behavioral goals and strategies, and reward and decision value outcome. Most of these streams converge on the cingulate domain, the main hub of the system. All of them are embedded within a larger eye–hand coordination network, from which they can be selectively set in motion by task demands.

Coupling between One-Dimensional Networks Reconciles Conflicting Dynamics in LIP and Reveals Its Recurrent Circuitry

Little is known about the internal circuitry of the primate lateral intraparietal area (LIP). During two versions of a delayed-saccade task, we found radically different network dynamics beneath similar population average firing patterns. When neurons are not influenced by stimuli outside their receptive fields (RFs), dynamics of the high-dimensional LIP network during slowly varying activity lie predominantly in one multi-neuronal dimension, as described previously. However, when activity is suppressed by stimuli outside the RF, slow LIP dynamics markedly deviate from a single dimension. The conflicting results can be reconciled if two LIP local networks, each underlying an RF location and dominated by a single multi-neuronal activity pattern, are suppressively coupled to each other. These results demonstrate the low dimensionality of slow LIP local dynamics, and suggest that LIP local networks encoding the attentional and movement priority of competing visual locations actively suppress one another.

Striatal Activity and Reward Relativity: Neural Signals Encoding Dynamic Outcome Valuation

The striatum is a key brain region involved in reward processing. Striatal activity has been linked to encoding reward magnitude and integrating diverse reward outcome information. Recent work has supported the involvement of striatum in the valuation of outcomes. The present work extends this idea by examining striatal activity during dynamic shifts in value that include different levels and directions of magnitude disparity. A novel task was used to produce diverse relative reward effects on a chain of instrumental action. Rats (Rattus norvegicus) were trained to respond to cues associated with specific outcomes varying by food pellet magnitude. Animals were exposed to single-outcome sessions followed by mixed-outcome sessions, and neural activity was compared among identical outcome trials from the different behavioral contexts. Results recording striatal activity show that neural responses to different task elements reflect incentive contrast as well as other relative effects that involve generalization between outcomes or possible influences of outcome variety. The activity that was most prevalent was linked to food consumption and post-food consumption periods. Relative encoding was sensitive to magnitude disparity. A within-session analysis showed strong contrast effects that were dependent upon the outcome received in the immediately preceding trial. Significantly higher numbers of responses were found in ventral striatum linked to relative outcome effects. Our results support the idea that relative value can incorporate diverse relationships, including comparisons from specific individual outcomes to general behavioral contexts. The striatum contains these diverse relative processes, possibly enabling both a higher information yield concerning value shifts and a greater behavioral flexibility.

A Bayesian model of context-sensitive value attribution

Substantial evidence indicates that incentive value depends on an anticipation of rewards within a given context. However, the computations underlying this context sensitivity remain unknown. To address this question, we introduce a normative (Bayesian) account of how rewards map to incentive values. This assumes that the brain inverts a model of how rewards are generated. Key features of our account include (i) an influence of prior beliefs about the context in which rewards are delivered (weighted by their reliability in a Bayes-optimal fashion), (ii) the notion that incentive values correspond to precision-weighted prediction errors, (iii) and contextual information unfolding at different hierarchical levels. This formulation implies that incentive value is intrinsically context-dependent. We provide empirical support for this model by showing that incentive value is influenced by context variability and by hierarchically nested contexts. The perspective we introduce generates new empirical predictions that might help explaining psychopathologies, such as addiction.

DOI:http://dx.doi.org/10.7554/eLife.16127.001

Neuronal Mechanisms of Visual Attention

Advances on several fronts have refined our understanding of the neuronal mechanisms of attention. This review focuses on recent progress in understanding visual attention through single-neuron recordings made in behaving subjects. Simultaneous recordings from populations of individual cells have shown that attention is associated with changes in the correlated firing of neurons that can enhance the quality of sensory representations. Other work has shown that sensory normalization mechanisms are important for explaining many aspects of how visual representations change with attention, and these mechanisms must be taken into account when evaluating attention-related neuronal modulations. Studies comparing different brain structures suggest that attention is composed of several cognitive processes, which might be controlled by different brain regions. Collectively, these and other recent findings provide a clearer picture of how representations in the visual system change when attention shifts from one target to another.

Adaptive neural coding: from biological to behavioral decision-making

Empirical decision-making in diverse species deviates from the predictions of normative choice theory, but why such suboptimal behavior occurs is unknown. Here, we propose that deviations from optimality arise from biological decision mechanisms that have evolved to maximize choice performance within intrinsic biophysical constraints. Sensory processing utilizes specific computations such as divisive normalization to maximize information coding in constrained neural circuits, and recent evidence suggests that analogous computations operate in decision-related brain areas. These adaptive computations implement a relative value code that may explain the characteristic context-dependent nature of behavioral violations of classical normative theory. Examining decision-making at the computational level thus provides a crucial link between the architecture of biological decision circuits and the form of empirical choice behavior.

Organization and evolution of parieto-frontal processing streams in macaque monkeys and humans

The functional organization of the parieto-frontal system is crucial for understanding cognitive-motor behavior and provides the basis for interpreting the consequences of parietal lesions in humans from a neurobiological perspective. The parieto-frontal connectivity defines some main information streams that, rather than being devoted to restricted functions, underlie a rich behavioral repertoire. Surprisingly, from macaque to humans, evolution has added only a few, new functional streams, increasing however their complexity and encoding power. In fact, the characterization of the conduction times of parietal and frontal areas to different target structures has recently opened a new window on cortical dynamics, suggesting that evolution has amplified the probability of dynamic interactions between the nodes of the network, thanks to communication patterns based on temporally-dispersed conduction delays. This might allow the representation of sensory-motor signals within multiple neural assemblies and reference frames, as to optimize sensory-motor remapping within an action space characterized by different and more complex demands across evolution.

On the functions, mechanisms, and malfunctions of intracortical contextual modulation

A broad neuron-centric conception of contextual modulation is reviewed and re-assessed in the light of recent neurobiological studies of amplification, suppression, and synchronization. Behavioural and computational studies of perceptual and higher cognitive functions that depend on these processes are outlined, and evidence that those functions and their neuronal mechanisms are impaired in schizophrenia is summarized. Finally, we compare and assess the long-term biological functions of contextual modulation at the level of computational theory as formalized by the theories of coherent infomax and free energy reduction. We conclude that those theories, together with the many empirical findings reviewed, show how contextual modulation at the neuronal level enables the cortex to flexibly adapt the use of its knowledge to current circumstances by amplifying and grouping relevant activities and by suppressing irrelevant activities.