Fast burst fraction transients convey information independent of the firing rate

Abstract

Theories of attention and learning have hypothesized a central role for high-frequency bursting in cognitive functions, but experimental reports of burst-mediated representations in vivo have been limited. Here we used a novel demultiplexing approach to separate independent streams of information from considering neurons as having three possible states: silent, singlet- and burst-firing. We studied this ternary neural code in vivo while animals learned to behaviorally report direct electrical stimulation of the somatosensory cortex and found two acquired yet independent representations. One code, the event rate, represented the stimulus in a small fraction of cells and showed a small modulation upon detection errors. The other code, the burst fraction, correlated more globally with stimulation and more promptly responded to detection errors. Bursting modulation was potent and its time course evolved, even in cells that were considered unresponsive based on the firing rate. During the later stages of training, this modulation in bursting happened earlier, gradually aligning temporally with the representation in event rate. The alignment of bursting and event rate modulation sharpened firing rate coded representations, and was strongly associated behavioral accuracy. Thus a fine grain separation of spike timing patterns reveals two signals that accompany stimulus representations: an error signal that can be essential to guide learning and a sharpening signal that could enact top-down attention.

Silences, spikes and bursts: Three‐part knot of the neural code

“…Consistently, Takahashi et al (2016) found that cells responding with a higher burst fraction were more likely to be predictive of the animal's perceptual report than cells with a lower burst fraction. Averaging over a population of cells, Naud et al (2022) showed that micro-stimulation induced increases in burst fraction over a population of cells. This change in burstiness adapted over multiple days, coming earlier as the animal became better at the detection task.…”

“…Furthermore, if bursting controls plasticity by providing a unit-specific representation of error Payeur et al (2021), one would predict that the relevant period for this instructive signal to act would come after any cues are given and after reward delivery or lack thereof. An increase in burstiness has been shown to encode errors with a delay of about 1 s with respect to the cue (Naud et al, 2022), suggesting that second-long eligibility traces similar to hippocampus may be at work in cortex. Consistently silencing pyramidal cell activity during the reward period but after the sensory cue has been shown to alter learning of a sensory association task (Ford et al, 2022).…”

“…The timing of burst occurrence shows much variability. Burst durations are also variable and appear mostly exponentially distributed (Wang et al., 2020; Naud et al., 2022). On average, 10%–35% of action potentials are in a burst, a rate that can be modulated by attention (Compte et al., 2000; Anderson et al., 2013; Womelsdorf et al., 2014) and fine task structure (Naud et al., 2022) by as much as 30%.…”

“…Burst durations are also variable and appear mostly exponentially distributed (Wang et al., 2020; Naud et al., 2022). On average, 10%–35% of action potentials are in a burst, a rate that can be modulated by attention (Compte et al., 2000; Anderson et al., 2013; Womelsdorf et al., 2014) and fine task structure (Naud et al., 2022) by as much as 30%. Modulations in burst probability can often be independent from modulations in firing rate (Naud et al., 2022).…”

“…On average, 10%–35% of action potentials are in a burst, a rate that can be modulated by attention (Compte et al., 2000; Anderson et al., 2013; Womelsdorf et al., 2014) and fine task structure (Naud et al., 2022) by as much as 30%. Modulations in burst probability can often be independent from modulations in firing rate (Naud et al., 2022). These in vivo observations suggest that the conjunctive burst code used by the cortex follows a rate code.…”