Neural representation of spatial information in the primate somatosensory system

The long term goal of this research is to understand the neural representation of tactile stimuli along the somatosensory pathway. To that end, this thesis consists of the systematic evaluation of models that can capture relevant aspects of neural responses at varying levels of abstraction. There are three studies:; The first, main study is the development of a continuum mechanical model for mechanotransduction to characterize the chain of events that culminates in the afferent response. The resulting model reliably predicts afferent responses to indented stimuli from quantities closely related to the local membrane stretch. The close agreement between model predictions and data supports the hypothesis that local membrane stretch in the skin drives mechanotransduction. The model is a powerful tool to evaluate hypotheses regarding the relationship between the peripheral neural representation and tactile perception.; The second study concerns the representation of spatial form in peripheral and cortical neurons, as characterized by their spatiotemporal receptive fields obtained using random dot stimuli. The main finding is that the representation of spatial form is transformed successively from peripheral afferents to cortical areas 3b and 1, through: (1) a progressive decrease in response linearity; (2) a progressive increase in inhibition in cortex; specifically, the proportion of surround inhibition increases from periphery to area 3b to area 1, indicating an increasingly complex processing of spatial form; and (3) a strikingly constant firing rate as a function of indentation density in cortex, compared to the logarithmic dependence of firing rates on indentation density in the periphery.; In another, related study, we design a novel set of stimuli that can be used to estimate a second spatiotemporal receptive field for every neuron, which is expected to be similar to the one obtained using random dots if the underlying neural response is linear. As a result, the difference between the two receptive fields provides a graphical depiction of the underlying response non-linearity. In cortex, excitatory regions in the two receptive fields were similar, whereas there was a considerable difference between inhibitory regions. Therefore, inhibitory sub-regions are likely less linear than their excitatory counterparts.; The third study is the detailed analysis, using a biophysical model for cortical neurons, to account for two types of data. First, the poisson-like discharge of a biophysical model for cortical neurons is rigorously reduced to a simple random walk; this result strengthens the intuitive link between the poisson-like discharge of cortical neurons and the membrane potential fluctuations that take place below the spiking threshold. Second, data from somatosensory cortex (SII) was analyzed to find a dynamic change in the population neuronal gain during a shift in attention, while preserving a constant discharge variability. Six mechanisms were identified in the biophysical model that might account for the observed attentional modulation. Of these, two mechanisms produced modulations consistent with the observed data: firing threshold and firing rate adaptation. In contrast, mechanisms such as changes in synchrony among the inputs produce firing rate modulations that are inconsistent with the observed data.