The sensory code within sense of time

Sensory experiences are accompanied by the perception of the passage of time; a cell phone vibration, for instance, is sensed as brief or long. The neuronal mechanisms underlying the perception of elapsed time remain unknown1. Recent work agrees on a role for cortical processing networks2,3, however the causal function of sensory cortex in time perception has not yet been specified. We hypothesize that the mechanisms for time perception are embedded within primary sensory cortex and are thus governed by the basic rules of sensory coding. By recording and optogenetically modulating neuronal activity in rat vibrissal somatosensory cortex, we find that the percept of stimulus duration is dilated and compressed by optogenetic excitation and inhibition, respectively, during stimulus delivery. A second set of rats judged the intensity of tactile stimuli; here, optogenetic excitation amplified the intensity percept, demonstrating sensory cortex to be the common gateway to both time and stimulus feature processing. The coding algorithms for sensory features are well established4–10. Guided by these algorithms, we formulated a 3–stage model beginning with the membrane currents evoked by vibrissal and optogenetic drive and culminating in the representation of perceived time; this model successfully replicated rats′ choices. Our finding that stimulus coding is intrinsic to sense of time disagrees with dedicated pacemaker-accumulator operation models11–13, where sensory input acts only to trigger the onset and offset of the timekeeping process. Time perception is thus as deeply intermeshed within the sensory processing pathway as is the sense of touch itself14,15 and can now be treated through the computational language of sensory coding. The model presented here readily generalizes to humans14,16 and opens up new approaches to understanding the time misperception at the core of numerous neurological conditions17,18.

Attentional dynamics of efficient codes

Top-down attention is hypothesized to dynamically allocate limited neural resources to task-relevant computations. According to this view, sensory neurons are driven not only by stimuli but also by feedback signals from higher brain areas that adapt the sensory code to the goals of the organism and its belief about the state of the environment. Here we formalize this view by optimizing a model of population coding in the visual cortex for maximally accurate perceptual inference at minimal activity cost. The resulting optimality predictions reproduce measured properties of attentional modulation in the visual system and generate novel hypotheses about the functional role of top-down feedback, response variability, and noise correlations. Our results suggest that a range of seemingly disparate attentional phenomena can be derived from a general theory combining probabilistic inference with efficient coding in a dynamic environment.

From an invariant sensory code to a perceptual categorical code in secondary somatosensory cortex

A crucial role of cortical networks is the conversion of sensory inputs into perception. In the cortical somatosensory network, neurons of the primary somatosensory cortex (S1) show invariant sensory responses, while frontal lobe neuronal activity correlates with the animal’s perceptual behavior. Here, we report that in the secondary somatosensory cortex (S2), neurons with invariant sensory responses coexist with neurons whose responses correlate with perceptual behavior. Importantly, the vast majority of the neurons fall along a continuum of combined sensory and categorical dynamics. These distinct neural responses exhibit analogous timescales of intrinsic fluctuations, suggesting that they belong to the same hierarchical processing stage. Furthermore, during a non-demanding control task, the sensory responses remained unaltered while perceptual responses vanished. Sensory information increased and categorical information diminished during this control task, suggesting that processing depended on the task context. Conclusively, S2 neurons exhibit intriguing dynamics that are intermediate between S1 and frontal lobe.

From an invariant sensory code to a perceptual categorical code in second somatosensory cortex

A crucial role of cortical networks is the conversion of sensory inputs into perception. In the cortical somatosensory network, neurons of the primary somatosensory cortex (S1) show invariant sensory responses, while frontal lobe neuron responses correlate with the animal’s perceptual behavior. But, where in the cortical somatosensory network are the sensory inputs transformed into perceptual behavior? Here, we report that in the secondary somatosensory cortex (S2), neurons with invariant sensory responses coexist with neurons whose responses correlate with the animal’s perceptual behavior. These distinct neural responses exhibit analogous timescales of intrinsic fluctuations, suggesting that they belong to the same hierarchical processing stage. Furthermore, during a non-demanding control task, the sensory responses remained unaltered while perceptual responses vanished. Conclusively, the S2 population responses exhibit intermediate dynamics between S1 and frontal lobe neurons. These results suggest that the conversion of touch into perception crucially depends on S2.

A multi-sensory code for emotional arousal

People express emotion using their voice, face and movement, as well as through abstract forms as in art, architecture and music. The structure of these expressions often seems intuitively linked to its meaning: romantic poetry is written in flowery curlicues, while the logos of death metal bands use spiky script. Here, we show that these associations are universally understood because they are signalled using a multi-sensory code for emotional arousal. Specifically, variation in the central tendency of the frequency spectrum of a stimulus—its spectral centroid—is used by signal senders to express emotional arousal, and by signal receivers to make emotional arousal judgements. We show that this code is used across sounds, shapes, speech and human body movements, providing a strong multi-sensory signal that can be used to efficiently estimate an agent's level of emotional arousal.

A multi-sensory code for emotional arousal

People express emotion using their voice, face and movement, as well as through abstract forms as in art, architecture and music. The structure of these expressions often seems intuitively linked to its meaning: e.g., romantic poetry is written in flowery curlicues, while the logos of death metal bands use spiky script. Here we show that these associations are universally understood because they are signaled using a multi-sensory code for emotional arousal. Specifically, variation in the central tendency of the frequency spectrum of a stimulus—its spectral centroid—is used by signal senders to express emotional arousal, and by signal receivers to make emotional arousal judgments. We show that this code is used across sounds, shapes, speech, and human body movements, providing a strong multi-sensory signal that can be used to efficiently estimate an agent’s level of emotional arousal.

Coding of Stimulus Frequency by Latency in Thalamic Networks Through the Interplay of GABAB-Mediated Feedback and Stimulus Shape

A temporal sensory code occurs in posterior medial (POm) thalamus of the rat vibrissa system, where the latency for the spike rate to peak is observed to increase with increasing frequency of stimulation between 2 and 11 Hz. In contrast, the latency of the spike rate in the ventroposterior medial (VPm) thalamus is constant in this frequency range. We consider the hypothesis that two factors are essential for latency coding in the POm. The first is GABAB-mediated feedback inhibition from the reticular thalamic (Rt) nucleus, which provides delayed and prolonged input to thalamic structures. The second is sensory input that leads to an accelerating spike rate in brain stem nuclei. Essential aspects of the experimental observations are replicated by the analytical solution of a rate-based model with a minimal architecture that includes only the POm and Rt nuclei, i.e., an increase in stimulus frequency will increase the level of inhibitory output from Rt thalamus and lead to a longer latency in the activation of POm thalamus. This architecture, however, admits period-doubling at high levels of GABAB-mediated conductance. A full architecture that incorporates the VPm nucleus suppresses period-doubling. A clear match between the experimentally measured spike rates and the numerically calculated rates for the full model occurs when VPm thalamus receives stronger brain stem input and weaker GABAB-mediated inhibition than POm thalamus. Our analysis leads to the prediction that the latency code will disappear if GABAB-mediated transmission is blocked in POm thalamus or if the onset of sensory input is too abrupt. We suggest that GABAB-mediated inhibition is a substrate of temporal coding in normal brain function.