Noise enables multiplexed coding of the amplitude & frequency of periodic signals in mouse primary somatosensory cortex

Kamaleddin, Mohammad Amin1, 2; Prescott, Steven A. 1, 2, 3

1. Neurosciences and Mental Health, The Hospital for Sick Children, Toronto, Ontario, Canada; 2. Institute of Biomaterials & Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada; 3. Department of Physiology, University of Toronto, Toronto, Ontario, Canada

One of the greatest challenges facing neuroscience is to understand how sensory stimuli are encoded in the brain, resulting in sensory perception. Different spatial and temporal features of neuronal spiking represent sensory information about external stimuli to the brain; however, the extent to which this information is being used by the brain for making perceptual decisions is elusive. When a neuron responds to a sensory stimulus, two fundamental codes may transmit the information: 1) the spike rate defined as the total number of spikes normalized by time; and 2) the spike timing defined as a detailed, millisecond-scale temporal structure of the spike pattern.

For vibrotactile stimulation driven by periodic signals of 100-600 Hz, emerging evidence suggests that spike rate and timing can both be used. Notably, pyramidal neurons in primary somatosensory cortex (S1) fire at rates << 100 Hz; temporal features of the spikes might be combined with spike rate to convey information. However, the presence of background noise is a defining property of cortical neurons. Many studies demonstrated that pyramidal neurons in the cortex are constantly bombarded by synaptic inputs, which can be approximated as colored noise. On one hand, it has been shown that a small perturbation in the background activity could result in the variation in the membrane potential independently from the stimulus and disrupt spike timing. On the other hand, noise has been shown to improve response fidelity. Thus, the extent to which noise might play a beneficial vs. detrimental role in rate and temporal coding of spikes, especially in cortical neurons, is yet to be fully resolved.

We recorded from pyramidal neurons in mouse S1 slices in the whole-cell configuration. Using the dynamic clamp technique to simulate a high-conductance state with different noise levels, we demonstrated that spikes occur at a preferred phase of the cyclic stimulus, consistent with a temporal coding of stimulus frequency. Moreover, firing rate is positively correlated with stimulus amplitude, consistent with a rate coding. Indeed, spikes occur at a preferred phase of a stimulus cycle – the basis for temporal coding of frequency – but a different number of stimulus cycles is skipped between consecutive spikes depending on stimulus intensity – the basis for rate coding of stimulus intensity. We showed that spike rate and spike timing encode different features of the stimulus, consistent with multiplexed coding, in which noise plays an important role in sending the dual information to the brain, facilitating the invariant representation of stimulus intensity through rate coding, while not disrupting the temporal coding of stimulus frequency. Our findings suggest that noise is not only not detrimental to the temporal coding of stimulus frequency, noise is also beneficial for the rate coding of stimulus intensity.