Recoding of Sensory Information across the Retinothalamic Synapse

Abstract

The neural code that represents the world is transformed at each stage of a sensory pathway. These transformations enable downstream neurons to recode information they receive from earlier stages. Using the retinothalamic synapse as a model system, we developed a theoretical framework to identify stimulus features that are inherited, gained, or lost across stages. Specifically, we observed that thalamic spikes encode novel, emergent, temporal features not conveyed by single retinal spikes. Furthermore, we found that thalamic spikes are not only more informative than retinal ones, as expected, but also more independent. Next, we asked how thalamic spikes gain sensitivity to the emergent features. Explicitly, we found that the emergent features are encoded by retinal spike pairs and then recoded into independent thalamic spikes. Finally, we built a model of synaptic transmission that reproduced our observations. Thus, our results established a link between synaptic mechanisms and the recoding of sensory information.

Energy Efficiency of Neural Information Processing

“…In vivo patch recordings, in cell-attached or whole-cell mode, were made from adult cats anesthetized with propofol and sufenta [55] [56] [57] [58] and were digitized at 10 or 25 kHz.…”

“…However, during active vision, only a fraction of the spikes traveling along the optic nerve can successfully activate a given downstream LGN cell, leading to a 2-to 4-fold reduction in the total number of spikes across the retinothalamic synaptic connection [55]. According to Barlow's principles of efficient coding [52], a minimum number of impulses should be used to encode an information source, which implies that the retinal and thalamic neural codes cannot be equally efficient.…”

“…According to Barlow's principles of efficient coding [52], a minimum number of impulses should be used to encode an information source, which implies that the retinal and thalamic neural codes cannot be equally efficient. The retinothalamic synapse selectively relays only the most informative retinal spikes, thereby preserving the most important information [53] [54] [55].…”

“…Here, we resolve this apparent paradox by analyzing simultaneous RGC and LGN recordings [55] with the Berger-Levy energy efficiency theory of neural computation and communication [5] [6]. Our analysis suggests that the retinal spike code, despite its relative inefficiency compared to the thalamic spike code, is an energy-efficient substrate for generating the thalamic code since it maximizes the information transmitted to the cortex per unit of expended energy.…”

“…The B-L theory predicts optimal input and output probability distributions that maximize the bits of information a neuron conveys to its efferent targets per joule of energy it expends for action potential (AP) generation, postsynaptic accumulation and basal metabolism. The theory implies (1) that the intervals between consecutive output spikes must be distributed according to a gamma distribution with shape and scale parameters denoted by κ and b, respectively, and (2) the scaled reciprocal of the input intensity must be distributed according to a beta distribution with three free parameters, two of them being the aforementioned κ and b and the third being a threshold-related parameter, m. In order to quantitatively compare the B-L theory with experimental observations, the spike train statistics from simultaneous recordings of synaptically-connected RGCs and LGN cells [55] were analyzed here using statistical fitting methods (see Section 5.4). Figure 5.1 illustrates the experimental setup with electrophysiological recordings simultaneously measuring the spikes coming from the retina and the LGN while stimulating ganglion cells with retinal inputs [55].…”