Search

Sucrose’s sweet intensity is one attribute contributing to the overconsumption of high-energy palatable foods. However, it is not known how sucrose intensity is encoded and used to make perceptual decisions by neurons in taste-sensitive cortices. We trained rats in a sucrose intensity discrimination task and found that sucrose evoked a widespread response in neurons recorded in posterior-Insula (pIC), anterior-Insula (aIC), and Orbitofrontal cortex (OFC). Remarkably, only a few Intensity-selective neurons conveyed the most information about sucrose’s intensity, indicating that for sweetness the gustatory system uses a compact and distributed code. Sucrose intensity was encoded in both firing-rates and spike-timing. The pIC, aIC, and OFC neurons tracked movement direction, with OFC neurons yielding the most robust response. aIC and OFC neurons encoded the subject’s choices, whereas all three regions tracked reward omission. Overall, these multimodal areas provide a neural representation of perceived sucrose intensity, and of task-related information underlying perceptual decision-making., eLife digest Imagine you wake up in the morning, and you pour yourself and your loved one coffee. They like it with two sugars but you only with one. Our ability to distinguish different sweet intensities allows us to detect how much sugar is in the coffee. It also helps us to predict the amount of energy present in foods and if it is safe to ingest. We can experience the sweet quality because our tongue contains sweet taste receptor cells that are switched on by sugar. This activates neurons across our taste system in the brain. However, we do not completely understand how these areas represent the intensity of sugar. Previous studies have only ‘passively’ measured different sugar concentrations, either using anesthetized animals or behavioral tasks that do not involve decision-making other than licking. But to accurately evaluate how animals perceive the intensity, active decision-making is required, such us ‘reporting’ the perceived concentration of sugar. Fonseca et al. set out to answer this question by training rats in a new sweet intensity discrimination task, in which the rats had to move to the left or right to obtain water as a reward. This way, the animals could ‘indicate’ how sweet they perceived the sugar water to be. At the same time, recordings from the three brain areas involved in taste responses were taken (called the anterior and posterior insular cortices, and the orbitofrontal cortex) to measure how the sugar intensity is processed in the brain. The results showed that a small group of neurons within all three areas contained more information about the sugar intensity than other neurons, suggesting the taste system uses a compact and distributed code to represent its intensity. The information about sugar intensity was contained in both the number of nerve impulses and in the precise timing with which these neurons fired. Many drinks and high-energy foods often contain large quantities of sugar, and their overconsumption contributes to the worldwide problems of obesity and its associated diseases. Therefore, a better understanding of the neurons that code information about the intensity of sugar could be a starting point for other studies to pinpoint the connections and areas in the brain involved in our irremediable attraction for sugar.

To prepare timely motor actions, we constantly predict future events. Regularly repeating events are often perceived as a rhythm to which we can readily synchronize our movements, just as in dancing to music. However, the neuronal mechanisms underlying the capacity to encode and maintain rhythms are not understood. We trained nonhuman primates to maintain the rhythm of a visual metronome of diverse tempos and recorded neural activity in the supplementary motor area (SMA). SMA exhibited rhythmic bursts of gamma band (30–40 Hz) reflecting an internal tempo that matched the extinguished visual metronome. Moreover, gamma amplitude increased throughout the trial, providing an estimate of total elapsed time. Notably, the timing of gamma bursts and firing rate modulations allowed predicting whether monkeys were ahead or behind the correct tempo. Our results indicate that SMA uses dynamic motor plans to encode a metronome for rhythms and a stopwatch for total elapsed time., eLife digest A catchy tune on the radio, and suddenly we are tapping our foot and moving our bodies to the rhythm of the music. We can follow a beat because our motor neurons, the nerve cells that control movements, work together in circuits. During actions that require precise timing – such as dancing to a rhythm – the motor neurons within these circuits increase and decrease their activity in complex patterns. But recent evidence shows that these motor neuron circuits also ‘switch on’ simply when we perceive a rhythm, even if we do not move to it. In fact, just imagining a rhythm triggers the same symphony of electrical activity in the brain. How do motor neurons generate coordinated patterns of activity without movement or even an external stimulus? Cadena-Valencia et al. set out to answer this question by training monkeys to follow a rhythm. The animals learned to track a dot that appeared alternately on the left and right sides of a touchscreen with a regular tempo. After a few repeats, the dot disappeared. The monkeys then had to continue mentally tracking where the dot would have been. A group of neurons in a brain region called the supplementary motor area synchronized their activity with the dot. Whenever the dot was due to appear, the neurons in the area showed a burst of rapid firing. These spikes of activity, called gamma bursts, helped the motor neurons to communicate with one another within their circuits. The gamma bursts thus acted as an internal metronome, making it easier for the monkeys to follow the rhythm. These results should be a starting point for other studies to pinpoint exactly where and how this rhythmic activity arises, and how the brain uses gamma bursts to synchronize our movements to a tempo.

Artificially activating neurons in the cortex can make a tetraplegic patient feel naturalistic sensations of skin pressure and arm movement.

Learning to associate unambiguous sensory cues with rewarded choices is known to be mediated by dopamine (DA) neurons. However, little is known about how these neurons behave when choices rely on uncertain reward-predicting stimuli. To study this issue we reanalyzed DA recordings from monkeys engaged in the detection of weak tactile stimuli delivered at random times and formulated a reinforcement learning model based on belief states. Specifically, we investigated how the firing activity of DA neurons should behave if they were coding the error in the prediction of the total future reward when animals made decisions relying on uncertain sensory and temporal information. Our results show that the same signal that codes for reward prediction errors also codes the animal’s certainty about the presence of the stimulus and the temporal expectation of sensory cues.

Timing is a fundamental variable for behavior. However, the mechanisms allowing human and non-human primates to synchronize their actions with periodic events are not yet completely understood. Here we characterize the ability of rhesus monkeys and humans to perceive and maintain rhythms of different paces in the absence of sensory cues or motor actions. In our rhythm task subjects had to observe and then internally follow a visual stimulus that periodically changed its location along a circular perimeter. Crucially, they had to maintain this visuospatial tempo in the absence of movements. Our results show that the probability of remaining in synchrony with the rhythm decreased, and the variability in the timing estimates increased, as a function of elapsed time, and these trends were well described by the generalized law of Weber. Additionally, the pattern of errors shows that human subjects tended to lag behind fast rhythms and to get ahead of slow ones, suggesting that a mean tempo might be incorporated as prior information. Overall, our results demonstrate that rhythm perception and maintenance are cognitive abilities that we share with rhesus monkeys, and these abilities do not depend on overt motor commands.

The visual system classifies objects into categories, and distinct populations of neurons within the temporal lobe respond preferentially to objects of a given perceptual category. We can also classify the objects we recognize with the sense of touch, but less is known about the neuronal correlates underlying this cognitive function. To address this question, we performed a multivariate pattern analysis (MVPA) of functional magnetic resonance imagining (fMRI) activity to identify the cortical areas that can be used to decode the category of objects explored with the hand. We observed that tactile object category can be decoded from the activity patterns of somatosensory and parietal areas. Importantly, we found that categories can also be decoded from the lateral occipital complex (LOC), which is a multimodal region known to be related to the representation of object shape. Furthermore, a hyperalignment analysis showed that activity patterns are similar across subjects. Our results thus indicate that tactile object recognition generates category-specific patterns of activity in a multisensory area known to encode objects, and that these patterns have a similar functional organization across individuals.

Recent studies that combined psychophysical/neurophysiological experiments [de Lafuente V, Romo R (2005) Nat Neurosci 8:1698–1703] analyzed the responses from single neurons, recorded in several cortical areas of parietal and frontal lobes, while trained monkeys reported the presence or absence of a mechanical vibration of varying amplitude applied to skin of one fingertip. The analysis showed that the activity of primary somatosensory cortex neurons covaried with the stimulus strength but did not covary with the animal's perceptual reports. In contrast, the activity of medial premotor cortex (MPC) neurons did not covary with the stimulus strength but did covary with the animal's perceptual reports. Here, we address the question of how perceptual detection is computed in MPC. In particular, we regard perceptual detection as a bistable neurodynamical phenomenon reflected in the activity of MPC. We show that the activity of MPC is consistent with a decision-making-like scenario of fluctuation-driven computation that causes a probabilistic transition between two bistable states, one corresponding to the case in which the monkey detects the sensory input, the other corresponding to the case in which the monkey does not. Moreover, the high variability activity of MPC neurons both within and between trials reflects stochastic fluctuations that may play a crucial role in the monkey's probabilistic perceptual reports.

Shadlen et al.’s Comment focuses on extrapolations of our results that were not implied or asserted in our Report. They discuss alternate analyses of average firing rates in other tasks, the relationship between neural activity and behavior, and possible extensions of the standard models we examined. Although interesting to contemplate, these points are not germane to the findings of our Report: that stepping dynamics provided a better statistical description of lateral intraparietal area spike trains than diffusion-to-bound dynamics for a majority of neurons.

When a sensory stimulus is presented, many cortical areas are activated, but how does the representation of a sensory stimulus evolve in time and across cortical areas during a perceptual judgment? We investigated this question by analyzing the responses from single neurons, recorded in several cortical areas of parietal and frontal lobes, while trained monkeys reported the presence or absence of a mechanical vibration of varying amplitude applied to the skin of one fingertip. Here we show that the strength of the covariations between neuronal activity and perceptual judgments progressively increases across cortical areas as the activity is transmitted from the primary somatosensory cortex to the premotor areas of the frontal lobe. This finding suggests that the neuronal correlates of subjective sensory experience gradually build up across somatosensory areas of the parietal lobe and premotor cortices of the frontal lobe.

When a near-threshold stimulus is presented, a sensory percept may or may not be produced. The unpredictable outcome of such perceptual judgment is believed to be determined by the activity of neurons in early sensory cortex. We analyzed the responses of neurons in primary somatosensory cortex, recorded while monkeys judged the presence or absence of threshold stimuli. We found that these responses did not covary with the monkeys' perceptual reports. In contrast, the activity of frontal lobe neurons did covary with trial-by-trial judgments. Further control and microstimulation experiments indicated that frontal lobe neurons are closely related to the monkeys' subjective experiences during sensory detection.

Key to understanding somatosensation is the form of how the mechanical stimuli are represented in the evoked neuronal activity of the brain. Here, we focus on studies that address the question of which components of the evoked neuronal activity in the somatosensory system represent the stimulus features. We review experiments that probe whether these neuronal representations are essential to somatosensation. We also discuss recent results that suggest how the somatosensory stimuli are represented in the brain during short-term memory. Finally, we review data that show the neuronal correlates of a decision during somatosensory perception.

Introduction to the Neurobiology of Interval Timing.- About the (non)scalar property for time perception.- Elucidating the internal structure of psychophysical timing performance in the sub-second and second range by utilizing confirmatory factor analysis.- Neurocomputational models of time perception.- Dedicated Clock/Timing-Circuit Theories of Time Perception and Timed Performance.- Neural Dynamics Based Timing in the Subsecond to Seconds Range.- Signs of timing in motor cortex during movement preparation and cue anticipation.- Neurophysiology of timing in the hundreds of milliseconds: multiple layers of neuronal clocks in the medial premotor areas.- The Olivo-Cerebellar System as a Neural Clock.- From duration and distance comparisons to goal encoding in prefrontal cortex.- Probing Interval Timing with Scalp-recorded Electroencephalography (EEG).- Searching for the Holy Grail: Temporally Informative Firing Patterns in the Rat.- Getting the timing right: experimental protocols for investigating time with functional neuroimaging and psychopharmacology.- Motor and Perceptual timing in Parkinson's disease.- Music Perception: Information Flow within the Human Auditory Cortices.- Perceiving temporal regularity in music: The role of auditory event-related potentials (ERPs) in probing beat perception.- Neural Mechanisms of Rhythm Perception: Present Findings and Future Directions.- Neural underpinnings of music: The polyrhythmic brain.

In perceptual decision-making tasks the activity of neurons in frontal and posterior parietal cortices covaries more with perceptual reports than with the physical properties of stimuli. This relationship is revealed when subjects have to make behavioral choices about weak or uncertain stimuli. If knowledge about stimulus onset time is available, decision making can be based on accumulation of sensory evidence. However, the time of stimulus onset or even its very presence is often ambiguous. By analyzing firing rates and correlated variability of frontal lobe neurons while monkeys perform a vibrotactile detection task, we show that behavioral outcomes are crucially affected by the state of cortical networks before stimulus onset times. The results suggest that sensory detection is partly due to a purely internal signal whereas the stimulus, if finally applied, adds a contribution to this initial processing later on. The probability to detect or miss the stimulus can thus be explained as the combined effect of this variable internal signal and the sensory evidence.

This paper reviews recent progress in understanding the functional roles of inhibitory interneurons in behaving animals and how they affect information processing in cortical microcircuits. Multiple studies have shown that the morphological subtypes of inhibitory cells show distinct electrophysiological properties, as well as different molecular and neurochemical identities, providing a large mosaic of inhibitory mechanisms for the dynamic processing of information in the cortex. However, it is only recently that some specific functions of different interneuronal subtypes have been described in behaving animals. In this regard, influential results have been obtained using the known differences of interneurons and pyramidal cells recorded extracellularly to dissociate the functional roles that these two classes of neurons may play in the cortical microcircuits during various behaviors. Neurons can be segregated into fast-spiking (FS) cells that show short action potentials, high discharge rates, and correspond to putative interneurons; and regular-spiking (RS) cells that show larger action potentials and correspond to pyramidal neurons. Using this classification strategy, it has been found that cortical inhibition is involved in sculpting the tuning to different stimulus or behavioral features across a wide variety of sensory, association, and motor areas. Recent studies have suggested that the increase in high-frequency synchronization during information processing and spatial attention may be mediated by FS activation. Finally, FS are active during motor planning and movement execution in different motor areas, supporting the notion that inhibitory interneurons are involved in shaping the motor command but not in gating the cortical output.

The brain constructs representations of objects and concepts based in sensory information combined with experience. This mental process, that we call perception, is the result of a chain of events consisting of phenomena such as detection, memory, discrimination, categorization and decision-making. Although the phenomenon of perception is not necessarily dependent on a given sensory modality (e.g. visual perception, auditory, tactile), single sensory models are indispensable for studying the neural mechanisms that generate it. The somatosensory system is a suitable model for studying the manner in which presentation of a single physical variable (e.g. vibration) triggers a perceptual process. Here, we discuss some recent studies in the somatosensory system that in our view, constitute a breakthrough to understanding decision making.

A fundamental problem in neurobiology is to understand how brain circuits represent sensory information and how such representations give rise to perception, memory and decision-making. We demonstrate that a sensory stimulus engages multiple areas of the cerebral cortex, including primary sensory, prefrontal, premotor and motor cortices. As information transverses the cortical circuits it shows progressively more relation to perception, memory and decision reports. In particular, we show how somatosensory areas on the parietal lobe generate a parameterized representation of a tactile stimulus. This representation is maintained in working memory by prefrontal and premotor areas of the frontal lobe. The presentation of a second stimulus, that monkeys are trained to compare with the first, generates decision-related activity reflecting which stimulus had the higher frequency. Importantly, decision-related activity is observed across several cortical circuits including prefrontal, premotor and parietal cortices. Sensory information is encoded by neuronal populations with opposite tuning, and suggests that a simple subtraction operation could be the underlying mechanism by which past and present sensory information is compared to generate perceptual decisions.

Midbrain dopamine (DA) neurons respond to sensory stimuli associated with future rewards. When reward is delivered probabilistically, DA neurons reflect this uncertainty by increasing their firing rates in a period between the sensory cue and reward delivery time. Probability of reward, however, has been externally conveyed by visual cues, and it is not known whether DA neurons would signal uncertainty arising internally. Here we show that DA neurons code the uncertainty associated with a perceptual judgment about the presence or absence of a vibrotactile stimulus. We observed that uncertainty modulates the activity elicited by a go cue instructing monkey subjects to communicate their decisions. That is, the same go cue generates different DA responses depending on the uncertainty level of a judgment made a few seconds before the go instruction. Easily detected suprathreshold stimuli elicit small DA responses, indicating that future reward will not be a surprising event. In contrast, the absence of a sensory stimulus generates large DA responses associated with uncertainty: was the stimulus truly absent, or did a low-amplitude vibration go undetected? In addition, the responses of DA neurons to the stimulus itself increase with vibration amplitude, but only when monkeys correctly detect its presence. This finding suggests that DA activity is not related to actual intensity but rather to perceived intensity. Therefore, in addition to their well-known role in reward prediction, DA neurons code subjective sensory experience and uncertainty arising internally from perceptual decisions.

Quite recently, it has become possible to use signals recorded simultaneously from large numbers of cortical neurons for real-time control. Such brain machine interfaces (BMIs) have allowed animal subjects and human patients to control the position of a computer cursor or robotic limb under the guidance of visual feedback. Although impressive, such approaches essentially ignore the dynamics of the musculoskeletal system, and they lack potentially critical somatosensory feedback. In this mini-symposium, we will initiate a discussion of systems that more nearly mimic the control of natural limb movement. The work that we will describe is based on fundamental observations of sensorimotor physiology that have inspired novel BMI approaches. We will focus on what we consider to be three of the most important new directions for BMI development related to the control of movement. (1) We will present alternative methods for building decoders, including structured, nonlinear models, the explicit incorporation of limb state information, and novel approaches to the development of decoders for paralyzed subjects unable to generate an output signal. (2) We will describe the real-time prediction of dynamical signals, including joint torque, force, and EMG, and the real-time control of physical plants with dynamics like that of the real limb. (3) We will discuss critical factors that must be considered to incorporate somatosensory feedback to the BMI user, including its potential benefits, the differing representations of sensation and perception across cortical areas, and the changes in the cortical representation of tactile events after spinal injury.

Imagine sitting next to a window in a restaurant atop a high building. You see people moving in the plaza below, and you instinctively direct your gaze in the direc- tion that most people are moving. This effortless action—identifying the net direc- tion of motion of a more-or-less randomly moving group of objects—involves cortical neurons from the middle temporal area (MT) of the brain. These neurons activate in response to visual objects moving in a par- ticular direction and inform the brain that an object is in motion 1 ,b ut understanding how this neuronal representation is used to causal relationships between neuronal fir- ing patterns and perception. Because MT neurons are very selective about the physi- cal attributes of the stimuli that activate them, the researchers first determined the appropriate combination of size, location, velocity and direction of the moving dots that optimally activated the neurons in the vicinity of the recording electrode. Once the best activating stimulus was found, the monkeys' task was designed such that the dots moved in the direction preferred by the neurons near the recording electrode in half of the trials, and in the opposite direc- tion in the other half. For example, if the recording electrode was near a group of neurons with upward direction preference, the stimulus set was designed such that monkeys had to decide whether the net motion of the dots was upward or down- ward. Thanks to this design, the authors could artificially activate neurons while monkeys performed the direction discrimi- nation task and add motion information that was either congruent or opposite to the direction of the moving dots. They could then evaluate how the addition of either type of information affected the monkeys' choice behavior. How this artificial addition of information affected the monkeys' behavior depended on the direction of the dots being presented with respect to the microstimulated neuron popu- lations. If the dots were moving in the direc- tion preferred by the microstimulated neurons, the authors found that the monkeys were substantially more likely to choose to make an eye movement in that direction. Additionally, the stimulus-viewing intervals (reaction times) were faster. This suggested that microstimulation was adding informa- tion to the stimulus, biasing the monkeys' behavioral responses in the direction pre- ferred by the stimulated neurons and causing the monkeys to reach faster decisions, pre- sumably due to a quicker accumulation of sensory evidence. These results elegantly confirmed the causal relationship between the activity of MT neurons and the percep- tion of motion direction 4 .

Summary Decisions emerge from the concerted activity of neuronal populations distributed across brain circuits. However, the analytical tools best suited to decode decision signals from neuronal populations remain unknown. Here we show that knowledge of correlated variability between pairs of cortical neurons allows perfect decoding of decisions from population firing rates. We recorded pairs of neurons from secondary somatosensory (S2) and premotor (PM) cortices while monkeys reported the presence or absence of a tactile stimulus. We found that while populations of S2 and sensory-like PM neurons are only partially correlated with behavior, those PM neurons active during a delay period preceding the motor report predict unequivocally the animal's decision report. Thus, a population rate code that optimally reveals a subject's perceptual decisions can be implemented just by knowing the correlations of PM neurons representing decision variables.

Decisions are often associated with a degree of certainty, or confidence — an estimate of the probability that the chosen option will be correct. Recent neurophysiological results suggest that the central processing of evidence leading to a perceptual decision also establishes a level of confidence. Here we provide a causal test of this hypothesis by electrically stimulating areas of the visual cortex involved in motion perception. Monkeys discriminated the direction of motion in a noisy display, and were sometimes allowed to opt out of the direction choice if their confidence was low. Microstimulation did not reduce overall confidence in the decision but instead altered confidence in a manner that mimicked a change in visual motion, plus a small increase in sensory noise. The results suggest that the same sensory neural signals support choice, reaction time and confidence in a decision, and that artificial manipulation of these signals preserves the quantitative relationship between accumulated evidence and confidence.

Sensory perception has traditionally been attributed to the activation of sensory cortices. However, by inducing an illusory perception of movement, Naito and colleagues show in this issue of Neuron that the illusory perception of movement is related to activation of primary motor cortex.

Simultaneous recordings of the firing activity of pairs of cortical neurons have shown that spike-count correlation coefficients (CCs) cover a wide range of values. According to recent theoretical and experimental work [1,2] recurrent cortical networks decorrelate neural activity producing very low CCs. However little is known about the origin of correlations and data analysis based on recordings of cortical activity of awake, behaving animals performing non-trivial tasks are scarce. In this study, we aim to understand the role of neural correlations in perceptual decision-making tasks and its relationship with the covariation between neural firing activity and behavior. We examine spike-count correlations obtained from pairs of simultaneously recorded premotor cortex (PC) neurons while trained monkeys performed a vibrotactile detection task in which the stimulus was often absent or weak, and the time of its application was variable [3,4]. By analyzing firing rates and correlated variability we show that behavioral outcomes are crucially affected by the state of cortical networks before stimulus onset times. Our results suggest that sensory detection is partly due to a purely internal signal whereas the stimulus, if finally applied, adds a contribution to this initial processing later on [5]. Noise correlations can be weak; their smallest values are attained at the end of the delay period of the task. Importantly, we found that small CCs are compatible with high firing rates. Although the firing rate in hit trials is higher than in correct rejections, the distributions of CCs over the population of pairs are similar, presenting mean values of 0.06. Moreover, the CCs do not covary with the geometrical mean of the firing rate of the pair. The observation that single neurons covary with the subject's response (characterized by the choice probability index, CP) is usually explained by the existence of variability correlations among the cells in a neuronal population [6]. Here we show a simple approximate expression that explicitly relates the population-averaged CP index and the CCs. This expression shows that the CP index is different from 0.5 when CCs evaluated using all trials differ from choice-conditioned correlations. Neurons could covary significantly with behavior even if the latter are very small. Thus, we show that there is no contradiction between the correlated activity required for the non-chance CP index and the small correlations produced by decorrelation in recurrent networks. Although the CP index is useful to study the role of single neurons in decision-making tasks, in reality, the decision is formed through the coordinated action of several pools of neurons. Hence, the relevant quantities to investigate the elaboration of the choice are population variables combining the activity of several pools. By extending the notion of CP from single neurons to neural pools, we defined the CPN index to quantify the amount of covariation with behavior of arbitrary linear combinations of oppositely-tuned neural pools. We found that pools of PC neurons exhibiting persistent activity become fully correlated with the subject's choice soon after stimulus onset and during the entire delay period of the task.

It was long suggested that the phasic activity of midbrain dopamine (DA) neurons codes the subject’s error in the prediction of reward [1]. Most of the experimental work in this field was done in the context of classical and instrumental conditioning. Recently, there has been an increasing interest to study the activity of DA neurons while the subject performs a decision-making task with the goal of obtaining reward at the end of the trial [2,3]. In an experiment in which monkeys have to decide about the presence or absence of a somatosensory stimulus, recordings of DA neurons have shown that the activity of these cells is modulated according to the trial type. The averaged activity in hit, miss, false alarm or correct rejection trials each presents a distinct temporal profile [3]. In particular, the neurons’ response to the go cue is correlated with the subject’s uncertainty about his choice. The signal of dopamine neurons in classical and instrumental conditioning has been explained in terms of the temporal-difference (TD) algorithm. However it is not clear whether and, if so, how reinforcement learning can account for the dopamine signals in complex decision-making tasks with noisy sensory information and temporal uncertainty of the relevant task events, as is the case in the detection task mentioned above [4]. We have developed an actor-critic model which deals with both these aspects of the problem. While an internal temporal representation keeps track of past relevant events, partial observability is accounted for by means of a Bayesian approach. The dopamine phasic activity predicted by the model matches the experimental data and the prediction of the psychometric curve is consistent with the animal performance. Furthermore, the model provides an interpretation of the condition-dependent dopamine response to the go cue instruction in terms of reward prediction error. Using Bayesian inference the model constructs an internal belief about the presence of the somatosensory stimulus. This belief reflects the confidence about the sensory perception and thus the value assigned to this perceptual judgment. The large belief in stimulus-present decisions represents a high degree of confidence in the sensory perception and a great expectation for future rewards. On the contrary, stimulus-absent choices reflect a small belief and consequently a larger uncertainty about the decision and the future reward. This computational description of belief agrees with the previous interpretation of the data [3] and describes well other experimental observations such as the dependence of DA neurons' signals on the stimulus amplitude. The model also predicts a decrease in dopamine activity before the go cue instruction, which is also observed in the data. We explain this decreasing tonic activity as an effect of the timing uncertainty. This is partly due to the task structure and partly generated from the limited temporal resolution of the stimulus representation, which creates subjective variability in the timing of events.