Cornelius Schwarz

The brain predicts action outcomes using interacting neuronal systems that integrate sensory feedback, movement, and rewards. In the mouse whisker system, we investigated whether two presumed predictive mechanisms jointly attenuate ascending tactile signals during movement. To this end, we employed an open-loop whisker-reach task that monitored facial nucleus motor activity and presented experimental tactile consequences while measuring tactile attenuation in the barrel cortex. Experimentally manipulating sensory consequences revealed two distinct systems. Sensory gating, a previously documented but functionally unclear mechanism of tactile attenuation, appeared graded but persisted during and after movement. In contrast, a second system was found to be temporally precise, generate suppression when stimuli are omitted, and capable of being learned within hundreds of trials. We label it state estimation as its characteristics—active suppression of expected tactile input—aligns well with the classical reafference principle, previously overlooked in the whisker system.

While implicit learning is thought to be responsible to generate conditioned responses (CRs) in Pavlovian delay paradigms, like delay eye blink conditioning (DEBC), explicit memory processes may be involved to learn the same contingencies, but store them for other purposes. Here, we investigated the engagement of the cortex in the supposedly non-cortical learning paradigm DEBC, using a single whisker deflection in head-fixed mice as the conditioned stimulus (CS). Spike responses in the primary somatosensory cortex (barrel cortex) showed two distinct changes. The first, a dip-peak pattern of firing-rate in response to CS, appeared many trials after the learned generation of CRs, while the second, a general decrement of firing in response to the CS, was tightly related to the time course of CR learning. The second effect matches previous findings using the cortex-dependent trace variant of eyeblink conditioning, while the first points to a learning-related cortical response that is independent from implicit learning and may well correspond to other types of learning. We hold this result as a first step to elucidate and disentangle neuronal underpinnings of multiple cooperative learning systems, commonly engaged even by the simplest forms of learning.

Perception and action are deeply intertwined processes that require the nervous system to distinguish between self-generated (reafferent) and externally generated (exafferent) sensory inputs. To maintain accurate perception during movement, the brain must attenuate predictable sensory consequences of its own actions while remaining sensitive to unexpected external events. Reafference attenuation is a temporally precise process that suppresses expected feedback, facilitating the detection of novel stimuli. This review examines reafference attenuation across species (rodents, nonhuman primates, and humans) and sensory systems (vestibular, auditory, and tactile). We also discuss sensory gating (or sensory suppression), a broader and often less selective mechanism that inhibits both self- and externally generated inputs. Although both mechanisms reduce sensory inflow during movement, they differ in function, specificity, and temporal dynamics, and despite growing insight into their underlying circuitry, important questions remain about their generality and implementation.

Successful sensorimotor behavior depends on anticipating when a tactile event will occur and translating it into rapid action, yet it remains unclear how action requirements shape temporal prediction in the human cortex. Here, we combined Magnetoencephalography (MEG), multivariate decoding and computational modeling to contrast temporally jittered finger stimulations under passive versus action-oriented conditions. We characterized a distributed frontal sensorimotor network that was linked to fast tactile-motor associations and predicted reaction-time variability. We found that trial-by-trial deviations from expected stimulus timing modulated behavior, with stronger frontocentral prediction-error encoding when an overt response was required. Temporal prediction errors were primarily encoded in a motor rather than sensory reference frame. These results reveal that action demands dynamically amplify temporal error signaling in sensorimotor circuits, in line with predictive coding frameworks postulating that the brain preferentially encodes task-relevant and goal-directed sensory prediction errors.

Can humans perceive the kinematic shape of very short tactile events? Textures impose such short events, frictional slips, which could be used for their discrimination. Using a pulse shape discrimination paradigm that enforces local analysis, we find that even the shape of isolated pulsatile skin deflections can be discriminated by humans. Providing an adaptive series of pulses, however, improves the performance, showing that the adaptive context plays an important role in instantaneous shape discrimination.

The brain is assumed to contain distinctive predictive systems ranging from sensorimotor to cognitive functions. Here we report the successful functional delineation of two different presumptive predictive systems on the neuronal level, state estimation (SE) and sensory gating (SG), which both attenuate sensory flow during movement. Studying neuronal sensorimotor responses throughout the depth of primary somatosensory cortex in mice trained on a whisker reach task, SE appeared as a learned attenuation of tactile signal flow, due to a suppressive predictive signal, precisely at the time of an experimental sensory consequence. In contrast, SG was observable during a much longer interval after onset of the motor command. Both phenomena are internal, presumptively predictive signals, as blockade of the reafference did not affect them. We speculate that SG may be related to cognitive signals monitoring goals of movements, while SE likely is the expression of the classical notion of the reafference principle.

In mammals several memory systems are responsible for learning and storage of associative memory. Even apparently simple behavioral tasks, like pavlovian conditioning, have been suggested to engage, for instance, implicit and explicit memory processes. Here, we used single-whisker tactile trace eyeblink conditioning (TTEBC) to investigate learning and its neuronal bases in the mouse barrel column, the primary neocortical tactile representation of one whisker. Behavioral analysis showed that conditioned responses (CRs) are spatially highly restricted; they generalize from the principal whisker only to its direct neighbors. Within the respective neural representation, the principal column and its direct neighbors, spike activity showed a learning-related spike rate suppression starting during the late phase of conditioning stimulus (CS) presentation that was sustained throughout the stimulus-free trace period (Trace). Trial-by-trial analysis showed that learning-related activity was independent from the generation of eyelid movements within a trial, and set in around the steepest part of the learning curve. Optogenetic silencing of responses and their learning-related changes during CS and Trace epochs blocked CR acquisition but not its recall after learning. Silencing during the Trace alone, which carried major parts of the learning-related changes, had no effect. In summary, we demonstrate specific barrel column spike rate plasticity during TTEBC that can be partially decoupled from the CR, the learned eye closure, a hallmark of implicit learning. Our results, thus, point to a possible role of the barrel column in contributing to other kinds of memory as well.

Tactile exploration often involves sequential touches interspersed with stimulus-free durations (e.g., the time during which the hand moves from one textured surface to the other). Whereas it is obvious that texture-related perceptual variables, irrespective of the encoding strategy, must be stored in memory for comparison, it is rather unclear which of those variables are held in memory. There are two established variables—“intensity” and “frequency”, which are “temporally global” variables because of the long stimulus integration interval required to average the signal or derive spectral components, respectively; on the other hand, a recently established third contender is the “temporally local” variable that codes for kinematic profiles of very short, suprathreshold events in the vibrotactile signal. Here, we present the first psychophysical evidence that temporally local variables can be stored in memory. To that end, we asked participants to detect changes in pulsatile indentation stimuli at their fingertips with and without a gap of 1 s between stimulus presentations. The stimuli either contained global variables alone (change of pulse rate), or a mix of local and global variables (change of pulse shape). We found, first, that humans are much better at detecting a change in stimuli when local variables are available rather than global ones alone—as evident by the fact that 21 compared to only 6 participants out of 25 yielded a valid psychophysical curve, respectively. Second, this observation persists even when there is a gap between the stimuli, implying local variables must be stored in memory. Third, an extensive array of relevant intensity definitions failed to explain participants’ performance in any consistent manner, which implies that perceptual decisions were less likely to be driven by intensity coding. Taken together, our results suggest that humans perform pulsatile change detection utilizing local pulse shape, and to a lesser degree global pulse rate, and that both parameters can be stored in memory.

The goal of cortical neuroprosthetics is to imprint sensory information as precisely as possible directly into cortical networks. Sensory processing, however, is dependent on the behavioral context. Therefore, a specific behavioral context may alter stimulation effects and, thus, perception. In this study, we reported how passive vs. active touch, i.e., the presence or absence of whisker movements, affects local field potential (LFP) responses to microstimulation in the barrel cortex in head-fixed behaving rats trained to move their whiskers voluntarily. The LFP responses to single-current pulses consisted of a short negative deflection corresponding to a volley of spike activity followed by a positive deflection lasting ~100 ms, corresponding to long-lasting suppression of spikes. Active touch had a characteristic effect on this response pattern. While the first phase including the negative peak remained stable, the later parts consisting of the positive peak were considerably suppressed. The stable phase varied systematically with the distance of the electrode from the stimulation site, pointing to saturation of neuronal responses to electrical stimulation in an intensity-dependent way. Our results suggest that modulatory effects known from normal sensory processing affect the response to cortical microstimulation as well. The network response to microstimulation is highly amenable to the behavioral state and must be considered for future approaches to imprint sensory signals into cortical circuits with neuroprostheses.