What, if anything, is the true neurophysiological significance of “rotational dynamics”?

Abstract

Back in 2012, Churchland and his colleagues proposed that "rotational dynamics", uncovered through linear transformations of multidimensional neuronal data, represents a fundamental type of neuronal population processing in a variety of organisms, from the isolated leech central nervous system to the primate motor cortex. Here, we evaluated this claim using Churchland's own data and simple simulations of neuronal responses. We observed that rotational patterns occurred in neuronal populations when (1) there was a temporal shift in peak firing rates exhibited by individual neurons, and (2) the temporal sequence of peak rates remained consistent across different experimental conditions. Provided that such a temporal order of peak firing rates existed, rotational patterns could be easily obtained using a rather arbitrary computer simulation of neural activity; no special provisions were needed for modeling motor cortical responses. Additionally, arbitrary traces, such as Lissajous curves, could be easily obtained from Churchland's data with multiple linear regression. While these observations suggest that "rotational dynamics" depict an orderly activation of different neurons in a population, we express doubt about Churchland et al.'s exaggerated assessment that this activity pattern is related to "an unexpected yet surprisingly simple structure in the population response" which "explains many of the confusing features of individual neural responses." Instead, we argue that their rotational dynamics approach provides little, if any, insight on the underlying neuronal mechanisms employed by neuronal ensembles to encode motor behaviors in any species.

Sequential and efficient neural-population coding of complex task information

“…Strikingly, this efficient-coding-like function did not seem to be implemented in a static way by dedicated neurons, but instead involved a sequence of different neurons across time. These sequential activity patterns corresponded to neural manifolds with a ring-like geometry 74 , and could be related to previous reports of lowdimensional manifolds corresponding to rotational neural dynamics 75 . We introduced a quantitative measure for the notion of sequentiality at the neural-population level, i.e.…”

Turning negative into positives! Exploiting ‘negative’ results in Brain–Machine Interface (BMI) research

“…Similarly, while many classification methods seem to be very efficient for offline BMI analysis, the methods actually used for online BMI are generally different, and often simpler [13]. Another example of the importance of discussing alternative interpretations is the case of the 'rotational dynamics' observed in intracranial recordings in pre-motor areas and their link to behaviour [16,17]. These examples highlight the fact that we need negative results in order to identify in which context a given method is the most suitable.…”

Cognitive computation using neural representations of time and space in the Laplace domain

“…This pre-movement firing was as if the entire sequential plan was quickly loaded into memory prior to movement initiation. Smooth reaching movements also result in sequentially activated cells in motor cortex (Lebedev et al, 2019). The similarity to stimulus-specific time cells suggests that these neural populations could code for an estimate of a function of sequential actions.…”

“…However, widely used datadriven approaches can be ill-suited to discover continua. For instance, individual reaching movements generate sequences of activity in motor cortex strikingly similar to sequences of time cells (Lebedev et al, 2019). These sequences can be readily understood as cells tiling a continuum, f ( * x).…”