Trial-to-trial variability in the responses of neurons carries information about stimulus location in the rat whisker thalamus

Abstract

From the perspective of neural coding, the considerable trial-totrial variability in the responses of neurons to sensory stimuli is puzzling. Trial-to-trial response variability is typically interpreted in terms of "noise" (i.e., it represents either intrinsic noise of the system or information unrelated to the stimuli). However, trial-totrial response variability can be considerably different across stimuli, suggesting that it could also provide an important contribution to the information conveyed by the neural responses about the stimuli. To test this hypothesis, we addressed the problem of discriminating stimulus location from the spike-count responses of neurons recorded in the ventro-postero-medial (VPM) nucleus of the thalamus in anesthetized rats. Using a recently developed information theory approach, we verified that differences between stimuli in the trial-to-trial spike-count variability of the responses provided an important contribution to the overall information carried by the neurons. In addition, we found that the relatively reliable (sub-Poisson) firing regime of our VPM neurons was not only more informative, but also more redundant between neurons compared with a more variable (Poisson) firing regime with the same total number of spikes. The typical increase in trial-to-trial response variability from the periphery to the cortex could therefore serve as a strategy to reduce redundancy between neurons and promote efficient sparse coding distributed in large populations of neurons. Overall, our data suggest that the trial-to-trial response variability plays a critical role in establishing the tradeoff between total information and redundancy between neurons in population codes.A major challenge for system neuroscience is to understand the basic elements of the neural code. Because single neurons typically respond to different stimuli with different average firing rates, possibly the simplest hypothesis is that neurons use a ratecoding scheme to represent sensory information (1, 2). However, average firing rates do not necessarily represent a complete description of the neural responses, partly because neurons can respond with a high degree of variability to different repetitions of the same stimulus (3). Trial-to-trial variability in the neural responses can be due to synaptic noise (4, 5) or can represent information that is not directly related to the stimulus, e.g., information about the state of the system (6-8). However, the relation between trial-to-trial variability in the responses of neurons and the information conveyed by those neurons remains unclear (9).The rate-coding hypothesis was originally formulated on the basis of classic works in peripheral nerves (10,11). Indeed, at the first stages of sensory processing, the trial-to-trial variability can be so low that neural responses can be almost deterministic (12-14), i.e., neurons respond with virtually the same number of spikes to different repetitions of the same stimulus. In the ideal deterministic regime, single-trial responses ar...

Sound-identity processing in early areas of the auditory ventral stream in the macaque

“…Obviously, to analyze correlations, multiple units must be recorded simultaneously with multiple electrodes, which limits the population size available for the analysis even if difficulties in using multielectrode arrays in cortical areas located within the lateral sulcus were overcome . Another approach that we decided not to use in this study is analysis of trial-to-trial variability of neural responses, which has been shown to play a role in cortical processing (e.g., Carandini 2004, Scaglione et al 2011, and to influence results of neuron-to-neuron correlation analysis (Ventura at al. 2005).…”

Spike count, spike timing and temporal information in the cortex of awake, freely moving rats

“…duration of the stimulus, see Methods) were randomly delivered to the mystacial pad via a whisker-pad stimulator (Moxon et al, 2007; Scaglione et al, 2011). Ninety-nine units were discriminated and the neuronal responses to the stimuli were separated into two groups based on whether the stimulus had been presented when the animal was sitting quietly or whisking.…”

“…Before the electrode array implantation surgery, a whisker-pad stimulator was implanted to be able to deliver electrical stimuli to the mystacial pad (Scaglione et al, 2011). The advantages of such stimulator are that: 1) the stimulator can be targeted to preferentially stimulate nerve fibers from a particular whisker or group of whiskers, providing a selectivity of afferent activation that was not available with the cuff electrode (Fanselow and Nicolelis, 1999) and 2) any contribution of direction selectively (Bale and Petersen, 2009; Simons and Carvell, 1989) on neuronal activity in response to the stimulation is minimized.…”

“…On the one hand, neurons can represent information by the total number of spikes emitted in a given time window (Adrian, 1926). With this code, generally referred to as spike count, the total information that can be carried by a neuron is primarily limited by its firing rate (Scaglione et al, 2011). On the other hand, neurons can use the time of arrival of individual spikes to represent information (Optican and Richmond, 1987; Bialek et al, 1991; Hopfield, 1995; deCharms and Merzenich, 1996; Victor and Purpura, 1996; de Ruyter van Steveninck et al, 1997; Borst and Theunissen, 1999; Reich et al, 2001; Chase and Young, 2006; Butts et al, 2007).…”

Brain-Machine Interfaces beyond Neuroprosthetics

“…In other words, instead of asking ourselves whether the brain uses spike-timing information based on analyses, we could require the brain to directly use spike timing and evaluate the outcome. Similar reasoning could be applied in a variety of classical neural coding problems, from the informational contribution of variability at the cellular level (Scaglione et al, 2011), to the role of correlations at the population level (Shamir, 2014), to the implications of theta-gamma oscillations (Lisman and Jensen, 2013;Bieri et al, 2014) and synchrony at higher network levels (Singer, 1999;Ratté et al, 2013). A proof of principle of this approach is already provided by the ability of monkeys to learn to generate gamma oscillations associated with spike synchrony in the motor cortex to control a BMI (Rouse et al, 2013;Engelhard et al, 2013; Figure 5A), the ability of rats to separately increase either firing rates and neural synchrony at cortico-hippocampal level to obtain a reward during operant conditioning (Sakurai and Takahashi, 2013) or, at non-invasive level, the ability of human subjects to modulate EEG power, frequency, phase, or even complexity in neurofeedback experiments (Brunner et al, 2006;Angelakis et al, 2007;Wang et al, 2011;So et al, 2014).…”