Highly resolved representations of dynamic haptic inputs in non-primary sensory cortical neuron representations

Abstract Recent studies suggest that the width and extent of cortical spread of sensory information appears to traditionally have been dramatically underestimated. Even neurons in primary visual cortex can carry detailed information about tactile inputs as shown in experiments using complex spatiotemporal electrotactile input patterns. Here we wanted to investigate if also detailed tactile information generated by complex haptic inputs is spread across the cortex. For this purpose, we used a vibrotactile haptic display with high temporal precision to stimulate the digits of the rat forepaw with sets of fine nuanced dynamic series of different frequencies. Multielectrode recordings were made from populations of neurons, almost all of which were located at substantial physiological distance from the direct sensory thalamocortical input. Regardless of location, we found large populations of neurons to separate the sets of nuanced haptic inputs, whereas no neuron had frequency-specific responses. Moreover, the combined activity of small populations of neighboring neurons supported time-resolved decoding of which stimulation pattern was currently playing, with pattern-specific decoding persisting for hundreds of milliseconds past stimulus offset (Figure 5). In descriptive terms this constitutes a short-term cortical memory of which input had just been presented, most evident in populations recorded closer to the direct sensory thalamocortical input. The use of light anesthesia and low-intensity inputs means that our data reflects the passive physiological cortical network propagation, as opposed to recent awake recordings where it can be unclear if the sensory input gradually snowballs into a behavioral response that then elicits a more indirect sensory cortical activation. Hence, highly specific information about haptic inputs is broadly distributed across the cortical network, present as locally decodable representations at cortical sites distant from the direct sensory thalamocortical input. Whether these locally decodable representations share a common code across regions is not addressed by the present single-insertion-per-animal design.