Reduced choice-related activity and correlated noise accompany perceptual deficits following unilateral vestibular lesion

Abstract

Signals from the bilateral vestibular labyrinths work in tandem to generate robust estimates of our motion and orientation in the world. The relative contributions of each labyrinth to behavior, as well as how the brain recovers after unilateral peripheral damage, have been characterized for motor reflexes, but never for perceptual functions. Here we measure perceptual deficits in a heading discrimination task following surgical ablation of the neurosensory epithelium in one labyrinth. We found large increases in heading discrimination thresholds and large perceptual biases at 1 wk postlesion. Repeated testing thereafter improved heading perception, but vestibular discrimination thresholds remained elevated 3 mo postlesion. Electrophysiological recordings from the contralateral vestibular and cerebellar nuclei revealed elevated neuronal discrimination thresholds, elevated neurometric-to-psychometric threshold ratios, and reduced trial-by-trial correlations with perceptual decisions ["choice probabilities" (CPs)]. The relationship between CP and neuronal threshold was shallower, but not significantly altered, suggesting that smaller CPs in lesioned animals could be largely attributable to greater neuronal thresholds. Simultaneous recordings from pairs of neurons revealed that correlated noise among neurons was also reduced following the lesion. Simulations of a simple pooling model, which takes into account the observed changes in tuning slope and correlated noise, qualitatively accounts for the elevated psychophysical thresholds and neurometric-to-psychometric ratios, as well as the decreased CPs. Thus, cross-labyrinthine interactions appear to play important roles in enhancing neuronal and perceptual sensitivity, strengthening interneuronal correlations, and facilitating correlations between neural activity and perceptual decisions.noise correlation | self-motion | decoding T he search for the neural basis of perception has fascinated neuroscientists for a long time. Traditional tools for assessing functional links between single neurons and perceptual decisions include quantification of trial-by-trial correlations between neuronal activity and perceptual reports for ambiguous stimuli (1). Such "choice probabilities" (CPs) have been reported in visual (2-9), parietal (10), and somatosensory (11) cortical areas. Computational studies have also shown that the extent of correlated noise in a pool of neurons profoundly affects . Until recently, these properties were largely considered the province of cerebral cortex.To better understand how peripheral sensory signals are processed centrally to generate perceptual decisions, Liu et al. (15) used a vestibular heading discrimination task to explore the relationships between neuronal sensitivity, CPs, and correlated noise in vestibular nucleus (VN) and cerebellar nucleus (CN) neurons. These areas receive afferent signals from the vestibular periphery and are interconnected with the thalamo-cortical system (16, 17), the midline vestibulocerebellum (18), the spinal cord ...

Decision-Related Activity in Sensory Neurons May Depend on the Columnar Architecture of Cerebral Cortex

“…However, this weak clustering for optic flow patterns could be merely a byproduct of the clustering for direction in MST (Celebrini and Newsome, 1994). If this were the case, these data are compatible with the observed pattern in cortex, whereas it may not hold for subcortical areas (Liu et al, 2013a;and see Discussion). Together, these results led us to hypothesize that decisionrelated activity is found only for those neurons that are organized in a map-like way for the task-relevant feature.…”

“…One reason why a columnar organization may be required is that theoretical (Shadlen et al, 1996) and empirical (Gu et al, 2008;Liu et al, 2013a) studies indicate a close relationship between decision-related activity and noise correlations. Unless a very small number (Ͻ10) of sensory neurons is used for a perceptual decision, the observation of choice probabilities requires that neurons that support the same decision are more strongly correlated than neurons supporting opposite decisions in the discrimination task.…”

Simple Learned Weighted Sums of Inferior Temporal Neuronal Firing Rates Accurately Predict Human Core Object Recognition Performance

“…Our comparison of nonhuman and human primates deviates from approaches that combine neural recording with behavioral testing in the same subjects (Britten et al, 1996;Luna et al, 2005;Cohen and Maunsell, 2011). It was motivated by our desire to get both high-fidelity behavioral and neuronal population data, a fruitful first-line strategy when a perceptual domain is poorly understood (Mountcastle et al, 1969;Johnson et al, 2002).…”

“…Specifically, our results argue that primates share a nonsemantic IT visual "feature" representation upon which semantic understanding can be learned, which constitutes a performance bottleneck in primate object recognition. This inference is agnostic as to how much of this feature representation is innate versus learned during the statistically shared postnatal experience of primates (Li and DiCarlo, 2008).…”

Neuronal thresholds and choice-related activity of otolith afferent fibers during heading perception

“…the neural population (38)(39)(40)(41)(42)(43)(44). When correlated noise is present, as is the case in the VN/CN (13,35), population thresholds may not decrease with the square root of the number of neurons and predictions based on the square root law could be dramatically inaccurate. Moreover, for a fixed neural pool size, population thresholds can also depend substantially on whether the decoder has full knowledge of the structure of noise correlations or not (42,45,46).…”