Bayes Optimality of Human Perception, Action and Learning: Behavioural and Neural Evidence (original) (raw)

Bayesian decision theory in sensorimotor control

Trends in cognitive sciences, 2006

Action selection is a fundamental decision process for us, and depends on the state of both our body and the environment. Because signals in our sensory and motor systems are corrupted by variability or noise, the nervous system needs to estimate these states. To select an optimal action these state estimates need to be combined with knowledge of the potential costs or rewards of different action outcomes. We review recent studies that have investigated the mechanisms used by the nervous system to solve such estimation and decision problems, which show that human behaviour is close to that predicted by Bayesian Decision Theory. This theory defines optimal behaviour in a world characterized by uncertainty, and provides a coherent way of describing sensorimotor processes.

Probabilistic optimization in the human perceptuo-motor system

The Journal of Physical Fitness and Sports Medicine, 2013

Despite the variability of internal and external environments, the human central nervous system (CNS) can generate precise and stable perception and motor behaviors. What mechanism enables this ability? Answering this question is one of the significant goals in the human sciences, including neuroscience, cognitive science, physical education and sports science. The Bayesian integration theory proposes that the CNS learns the prior distribution of a task and integrates it with sensory information to minimize the effect of sensory noise. In this article, we introduce psychophysical reports using motor timing and temporal order judgment (TOJ) tasks that support the Bayesian integration theory. Subsequently, we demonstrate the event-related potentials (ERPs) behind Bayesian integration that operates in somatosensory TOJ.

Neural basis of rule-based decisions with graded choice biases

According to an emerging view, decision-making and motor planning are tightly entangled at the level of neural processing. Choice is influenced not only by the values associated with different options, but also biased by other factors. Here we test the hypothesis that preliminary action planning can induce choice biases gradually and independently of objective value when planning overlaps with one of the potential action alternatives. Subjects performed center-out reaches obeying either a clockwise or counterclockwise cue-response rule in two tasks. In the probabilistic task, a pre-cue indicated the probability of each of the two potential rules to become valid. When the subsequent rulecue unambiguously indicated which of the pre-cued rules was actually valid (instructed trials), subjects responded faster to rules pre-cued with higher probability. When subjects were allowed to choose freely between two equally rewarded rules (choice trials) they chose the originally more likely rule more often and faster, despite the lack of an objective advantage in selecting this target. In the amount task, the pre-cue indicated the amount of potential reward associated with each rule. Subjects responded faster to rules pre-cued with higher reward amount in instructed trials of the amount task, equivalent to the more likely rule in the probabilistic task. Yet, in contrast, subjects showed hardly any choice bias and no increase in response speed in favor of the original high-reward target in the choice trials of the amount task. We conclude that free-choice behavior is robustly biased when predictability encourages the planning of one of the potential responses, while prior reward expectations without HUMAN PSYCHOPHYSICS 24 action planning do not induce such strong bias. Our results provide behavioral evidence for distinct contributions of expected value and action planning in decision-making and a tight interdependence of motor planning and action selection, supporting the idea that the underlying neural mechanisms overlap. We conclude that free-choice behavior is particularly biased when pre-cues allow the planning of one response over another alternative. Our results provide behavioral evidence for distinct contributions of expected value and action planning in decisionmaking process. The results also provided evidence for the tight interdependence of decision behavior and motor planning, thereby supporting the idea that the underlying neural mechanisms overlap.

Psychology and neurobiology of simple decisions

Trends in Neurosciences, 2004

Patterns of neural firing linked to eye movement decisions show that behavioral decisions are predicted by the differential firing rates of cells coding selected and nonselected stimulus alternatives. These results can be interpreted using models developed in mathematical psychology to model behavioral decisions. Current models assume that decisions are made by accumulating noisy stimulus information until sufficient information for a response is obtained. Here, the models, and the techniques used to test them against response-time distribution and accuracy data, are described. Such models provide a quantitative link between the time-course of behavioral decisions and the growth of stimulus information in neural firing data.

From perception to action: An economic model of brain processes

Games and Economic Behavior, 2012

We build on research from neurobiology to model the process through which the brain maps outside evidence into decisions. The sensory system encodes information through cell-firing. Cell-firing is measured against a threshold, and an action is triggered depending on whether the threshold is surpassed. The decision system modulates the threshold. We show that the (constrained) optimal threshold is set in a way that existing beliefs are likely to be confirmed. We then derive behavioral implications. Our mechanism can explain in a unified framework a number of 'anomalies' noted in psychology and economics: (i) belief anchoring (the order in which evidence is received affects beliefs and choices); (ii) polarization (individuals with opposite priors may polarize their opinions after receiving identical evidence); (iii) payoff-dependence of beliefs and (iv) belief disagreement (individuals with identical priors who receive the same evidence may end up with different posterior beliefs).

A Biased Bayesian Inference for Decision-Making and Cognitive Control

Frontiers in Neuroscience

Although classical decision-making studies have assumed that subjects behave in a Bayes-optimal way, the sub-optimality that causes biases in decision-making is currently under debate. Here, we propose a synthesis based on exponentially-biased Bayesian inference, including various decision-making and probability judgments with different bias levels. We arrange three major parameter estimation methods in a two-dimensional bias parameter space (prior and likelihood), of the biased Bayesian inference. Then, we discuss a neural implementation of the biased Bayesian inference on the basis of changes in weights in neural connections, which we regarded as a combination of leaky/unstable neural integrator and probabilistic population coding. Finally, we discuss mechanisms of cognitive control which may regulate the bias levels.

Neural Mechanisms of Human Decision-Making

Cognitive, Affective, & Behavioral Neuroscience, 2021

We present a theory and neural network model of the neural mechanisms underlying human decision-making. We propose a detailed model of the interaction between brain regions, under a proposer-predictor-actor-critic framework. This theory is based on detailed animal data and theories of action-selection. Those theories are adapted to serial operation to bridge levels of analysis and explain human decision-making. Task-relevant areas of cortex propose a candidate plan using fast, model-free, parallel neural computations. Other areas of cortex and medial temporal lobe can then predict likely outcomes of that plan in this situation. This optional prediction-(or model-) based computation can produce better accuracy and generalization, at the expense of speed. Next, linked regions of basal ganglia act to accept or reject the proposed plan based on its reward history in similar contexts. If that plan is rejected, the process repeats to consider a new option. The reward-prediction system acts as a critic to determine the value of the outcome relative to expectations and produce dopamine as a training signal for cortex and basal ganglia. By operating sequentially and hierarchically, the same mechanisms previously proposed for animal action-selection could explain the most complex human plans and decisions. We discuss explanations of model-based decisions, habitization, and risky behavior based on the computational model.

Neural correlates of prior probability processing in decision-making

International Journal of Psychophysiology, 2008

When selecting actions in response to noisy sensory stimuli, the brain can exploit prior knowledge of time constraints, stimulus discriminability and stimulus probability to hone the decision process. Although behavioral models typically explain such effects through adjustments to decision criteria only, the full range of underlying neural process adjustments remains to be established. Here, we draw on human neurophysiological signals reflecting decision formation to construct and constrain a multi-tiered model of prior-informed motion discrimination, in which a motor-independent representation of cumulative evidence feeds build-to-threshold motor signals that receive additional dynamic urgency and bias signal components. The neurally-informed model not only provides a superior quantitative fit to prior-biased behavior across three distinct task regimes (easy, time-pressured and weak evidence), but also reveals adjustments to evidence accumulation rate, urgency rate, and the timing of accumulation onset and motor execution which go undetected or are discrepant in more standard diffusionmodel analysis of behavior. .