Shared representation of licking directions in the preparatory population activities of anterior lateral motor cortical neurons in mice

Abstract

Anterior lateral motor cortex (ALM) in the mouse is involved in the formation and retention of future licking direction. The majority of ALM neurons reveal selectivity that characterizes differential firing rates of these neurons depending on future licking direction. Many researchers have found that interaction of ALM with other brain regions such as barrel cortex, medial motor cortex and thalamus are necessary for movement preparation in ALM. Also, recent studies have shown that motor cortical neurons of non-human primates and rodents covary on a low-dimensional shared space, where neural population activities are jointly represented, and the shared space is realigned through learning to achieve a goal. Although local and cross-regional circuits of ALM during movement preparation are well documented, presentation of selectivity in the firing activities of ALM neurons still remains elusive. In this study, we hypothesized that selectivity would be the result of the coordination of joint activity of ALM neurons represented in a low-dimensional shared space. To address this hypothesis, we analyzed the public data from the CRCNS repository. These data were collected from the experiment of a tactile decision making task. In this task, a pole initially touched the whisker of the mice for 1.3s informing the direction of the reward. After the pole was detached from the whisker, mice waited for 1.3s (delay period) and licked to the right if the pole had touched a posterior part or to the left if it had touched an anterior part of the whisker (response period). We used factor analysis to decompose the firing rates of ALM neurons in the delay period into shared signals, a portion of a total firing rate explained by covarying activities among neurons, and private signals, the remaining portion showing individual firing rates uncorrelated with each other. We calculated the selectivity and communality of individal neurons where communality refers to the amount of variance of firing rates explained by the factors. We then measured correlations between selectivity and communality for the hit trials (the mouse licked to the correct direction) or the error trials (the mouse licked to the incorrect direction). We found relatively strong positive correlations between selectivity and communality in 74% of the neurons that showed significant selectivity in the delay period (137 neurons, r2 =0.45, p<0.01). It demonstrates that neurons with higher communality in joint activity tended to have greater selectivity. These correlations became weaker in the error trials (r2 =0.04, p<0.01). Remarkably, there was a tendency that low communality in the hit trials was increased in the error trials whereas high communality became lower (t-test, p<0.01). It indicates that a specific group of neurons had to covary in the shared space to a certain degree for successful movement preparation. Our results may suggest that selectivity of ALM neurons could be coupled with coordinated activity of a neural population represented in the shared space.