BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A universal probabilistic spike count model reveals ongoing modula
 tion of neural variability in head direction cell activity in mice - David
  Liu\, Department of Engineering
DTSTART:20211027T160000Z
DTEND:20211027T170000Z
UID:TALK164299@talks.cam.ac.uk
CONTACT:Katharina Zuhlsdorff
DESCRIPTION:Neural responses are variable: even under identical experiment
 al conditions\, single neuron and population responses typically differ fr
 om trial to trial and across time. Recent work has demonstrated that this 
 variability has predictable structure\, can be modulated by sensory input 
 and behaviour\, and bears critical signatures of the underlying network dy
 namics and computations. However\, current methods for characterising neur
 al variability are primarily geared towards sensory coding in the laborato
 ry: they require trials with repeatable experimental stimuli and behaviour
 al covariates. In addition\, they make strong assumptions about the parame
 tric form of variability\, rely on assumption-free but data-inefficient hi
 stogram-based approaches\, or are altogether ill-suited for capturing vari
 ability modulation by covariates. Here we present a universal probabilisti
 c spike count model that eliminates these shortcomings. Our method uses sc
 alable Bayesian machine learning techniques to model arbitrary spike count
  distributions (SCDs) with flexible dependence on observed as well as late
 nt covariates. Without requiring repeatable trials\, it can flexibly captu
 re covariate-dependent joint SCDs\, and provide interpretable latent cause
 s underlying the statistical dependencies between neurons. We apply the mo
 del to recordings from a canonical non-sensory neural population: head dir
 ection cells in the mouse. We find that variability in these cells defies 
 a simple parametric relationship with mean spike count as assumed in stand
 ard models\, its modulation by external covariates can be comparably stron
 g to that of the mean firing rate\, and slow low-dimensional latent factor
 s explain away neural correlations. Our approach paves the way to understa
 nding the mechanisms and computations underlying neural variability under 
 naturalistic conditions\, beyond the realm of sensory coding with repeatab
 le stimuli.
LOCATION:Zoom
END:VEVENT
END:VCALENDAR