Attractors and flows in the neural dynamics of movement control

BAGLIETTO G.; STEFANO FERRAINA; DEL GIUDICE, PAOLO

Barcelona

Conferencia; 28th Annual Computational Neuroscience Meeting: CNS*2019; 2019

Density-based clustering (DBC) [1] provides efficient representationsof a multidimensional time series, allowing to cast it in the form of thesymbolic sequence of the labels identifying the cluster to which eachvector of instantaneous values belong. Such representation naturallylends itself to obtain compact descriptions of data from multichannelelectrophysiological recordings.We used DBC to analyze the spatio-temporal dynamics of dorsal pre-motor cortex in neuronal data recorded from two monkeys during a?countermanding? reaching task: the animal must perform a reachingmovement to a target on a screen (?no-stop trials?), unless an interven-ing stop signal prescribes to withhold the movement (?stop-trials?); no-stop (~70%) and stop trials (~30%) were randomly intermixed, and thestop signal occurred at variable times within the reaction time.Multi-unit activity (MUA) was extracted from signals recorded using a96-electrodes array. Performing DBC on the 96-dimensional MUA timeseries, we derived the corresponding discrete sequence of clusters?centroid.Through the joint analysis of such cluster sequences for no-stop andstop trials we show that reproducible cluster sequences are associ-ated with the completion of the motor plan in no-stop trials, and thatin stop trials the performance depends on the relative timing of suchstates and the arrival of the Stop signal.Besides, we show that a simple classifier can reliably predict the out-come of stop trials from the cluster sequence preceding the appear-ance of the stop signal, at the single-trial level.We also observe that, consistently with previous studies, the inter-trialvariability of MUA configurations typically collapses around the move-ment time, and has minima corresponding to other behavioral events(Go signal; Reward); comparing the time profile of MUA inter-trial vari-ability with the cluster sequences, we are led to ask whether the neu-ral dynamics underlying the clusters sequence can be interpreted asattractor hopping. For this purpose we analyze the flow in the MUAconfiguration space: for each trial, and each time, the measured MUAvalues identify a point in the 96-dimensional space, such that each trialcorresponds to a trajectory in this space, and a set of repeated trials toa bundle of trajectories, of which we can compute individual or aver-age properties. We measure quantities suited to discriminate betweena dynamics of convergence of the trajectories to a point attractor,from different flows in the MUA configuration space. We tentativelyconclude that convergent attractor relaxation dynamics (in attentivewait conditions, as before the Go or the Reward events) coexist withcoherent flows (associated with movement onset), in which low inter-trial variability of MUA configurations corresponds to a collapse in thedirections of velocities (with high magnitude of the latter), like the sys-tem entering a funnel.The ?delay task? (Go signal comes with a variable delay after the visualtarget), allows to further check our interpretation of specific MUA con-figurations (clusters) as being associated with the completion of themotor plan. Preliminary analysis shows that pre-movement-related MUA cluster sequences during delay trials are consistent with thosefrom other trial types, though their time course qualitatively differs inthe two monkeys, possibly reflecting different computational options.