Chronic large-scale recording with ultraflexible electrode arrays for studying neural codes and their stability

A central question in neuroscience is identifying neural codes that stably represent external variables across time. Using mice visual perception as the experimental paradigm, I focused on the debate between rate code versus temporal code based neural representation and depicted their differential contribution to the duality of neural representation stability and drift. Despite the pivotal distinction between spike counting based rate code and spike timing aware temporal codes, previous studies have yet to unveil the role of temporal code in long-term visual representation due to technical constraints. Past reports on drift in visual code over time predominantly relied on calcium imaging, which lacked the temporal resolution to capture fast-spiking dynamics and were further confounded by interferences such as photobleaching, leaving a gap in our comprehension of these complexities in neural code. While such investigation could have been carried out with electrophysiological recordings that resolves fast spiking dynamics, the scale and longevity necessary to study representation stability has not been achieved with conventional rigid electrodes. Our group overcomes these hurdles with large scale implantation of ultraflexible nanoelectronic threads (NETs) electrodes, which provide unprecedented longitudinal recordings across many neurons, while minimizing tissue-electrode interface instability. In this thesis, I a) established a platform to map visual response properties of neural units from > 1000 channels of ultraflexible electrodes. b) developed a method to track same units recorded by these ultraflexible electrode arrays. c) compared the stability of different neural codes by longitudinally tracking > 1000 single neuron units from 5 mice over 15 consecutive days from animals subjected to repeated, diverse visual stimuli every day. Our result reveals that considering the fast temporal dynamics of neuronal spikes (temporal code) enhances the stability of individual neuron tuning, neuronal population representation, and decoding accuracy compared to rate code. Thus, temporal coding, a mechanism that operates on the millisecond scale of neural communication, might be a fundamental principle that supports the consistency of sensory experiences, amidst the ever-changing brain states and synaptic strengths.