Direct observation of the neural computations underlying a single decision

Abstract

Neurobiological investigations of perceptual decision-making have furnished the first glimpse of a flexible cognitive process at the level of single neurons (Shadlen & Newsome, 1996; Shadlen & Kiani 2013). Neurons in the parietal and prefrontal cortex (Kim & Shadlen, 1999; Romo, Hernandez & Zainos, 2004; Hernandez, Zainos & Romo, 2002; Ding & Gold, 2012) are thought to represent the accumulation of noisy evidence, acquired over time, leading to a decision. Neural recordings averaged over many decisions have provided support for the deterministic rise in activity to a termination bound (Roitman & Shadlen, 2002). Critically, it is the unobserved stochastic component that is thought to confer variability in both choice and decision time (Gold & Shadlen, 2007). Here, we elucidate this stochastic, diffusion-like signal on individual decisions by recording simultaneously from hundreds of neurons in the lateral intraparietal cortex (LIP). We show that a small subset of these neurons, previously studied singly, represent a combination of deterministic drift and stochastic diffusion—the integral of noisy evidence—during perceptual decision making, and we provide direct support for the hypothesis that this diffusion signal is the quantity responsible for the variability in choice and reaction times. Neuronal state space and decoding analyses, applied to the whole population, also identify the drift diffusion signal. However, we show that the signal relies on the subset of neurons with response fields that overlap the choice targets. This parsimonious observation would escape detection by these powerful methods, absent a clear hypothesis.

A neural mechanism for terminating decisions

“…We show that in LIP, these averages belie drift-diffusion signals on single decisions (Fig. 3A; see Steinemann et al 2022). In stark contrast to LIP, the trial-averaged spike rates in SC belie non-saccadic bursts.…”

“…Until now, the mechanism for how this is achieved was unknown. We show that neurons in LIP represent the accumulation of noisy evidence on single decisions-the latent drift-diffusion process posited by evidence accumulation models (see also Steinemann et al, 2022). SC, which is reciprocally connected with LIP, performs a distinct computation.…”

“…It is only evident in the single trial averages. In an accompanying paper, we show that such single trial firing rates establish the decision variable that determines the choice and RT on each decision (see Steinemann et al, 2022, for details).…”

“…The preservation of upticks when SC is inactivated suggests that LIP does not depend on the portion of inactivated SC to generate the upticks. The upticks themselves are part of the decision variable in LIP-the rising phases of drift diffusion signals (Steinemann et al, 2022).…”

“…We conducted five recording sessions with the Neuropixels probe in Monkey M and three sessions in Monkey J, yielding 54-203 LIP neurons per session. These 8 sessions are also included in a companion paper focusing on the drift-diffusion signal in LIP (Steinemann et al, 2022). Neural data from these probes were recorded using SpikeGLX software and were synced post-hoc to behavioral data, task events, and SC activity recorded with an Omniplex system (Plexon).…”