Modeling statistical dependencies in multi-region spike train data

Keeley, Stephen; Zoltowski, David M.; Aoi, Mikio C.

doi:10.1016/j.conb.2020.11.005

Cited by 38 publications

(34 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our approach can also be extended in several ways to make it more useful to the neuroscience community. For example, replacing the spike count-based noise models with a point process model would provide higher temporal resolution (Duncker and Sahani, 2018), and facilitate inference of optimal temporal delays across neural populations (Lakshmanan et al, 2015) which will likely be useful as multi-region recordings become more prevalent in neuroscience (Keeley et al, 2020). Additionally, by substituting the linear kernel in p ( Y | X ) for an RBF kernel in Euclidean space (Wu et al, 2017) or on a non-Euclidean manifold (Jensen et al, 2020), we can recover scalable versions of recent GPLVM-based tools for neural data analyses with automatic relevance determination.…”

Section: Discussionmentioning

confidence: 99%

Scalable Bayesian GPFA with automatic relevance determination and discrete noise models

Jensen

Kao

Stone

et al. 2021

Preprint

View full text Add to dashboard Cite

Latent variable models are ubiquitous in the exploratory analysis of neural population recordings, where they allow researchers to summarize the activity of large populations of neurons in lower dimensional 'latent' spaces. Existing methods can generally be categorized into (i) Bayesian methods that facilitate flexible incorporation of prior knowledge and uncertainty estimation, but which typically do not scale to large datasets; and (ii) highly parameterized methods without explicit priors that scale better but often struggle in the low-data regime. Here, we bridge this gap by developing a fully Bayesian yet scalable version of Gaussian process factor analysis (bGPFA) which models neural data as arising from a set of inferred latent processes with a prior that encourages smoothness over time. Additionally, bGPFA uses automatic relevance determination to infer the dimensionality of neural activity directly from the training data during optimization. To enable the analysis of continuous recordings without trial structure, we introduce a novel variational inference strategy that scales near-linearly in time and also allows for non-Gaussian noise models more appropriate for electrophysiological recordings. We apply bGPFA to continuous recordings spanning 30 minutes with over 14 million data points from primate motor and somatosensory cortices during a self-paced reaching task. We show that neural activity progresses from an initial state at target onset to a reach-specific preparatory state well before movement onset. The distance between these initial and preparatory latent states is predictive of reaction times across reaches, suggesting that such preparatory dynamics have behavioral relevance despite the lack of externally imposed delay periods. Additionally, bGPFA discovers latent processes that evolve over slow timescales on the order of several seconds and contain complementary information about reaction time. These timescales are longer than those revealed by methods which focus on individual movement epochs and may reflect fluctuations in e.g. task engagement.

show abstract

Section: Discussionmentioning

confidence: 99%

Scalable Bayesian GPFA with automatic relevance determination and discrete noise models

Jensen

Kao

Stone

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We next further analyzed the peri-stimulus-time histograms (PSTHs) of A1 and MGBv during stimulus presentation. We examined how neuronal activity in A1 was related to that in MGBv, using regression-based approaches (Semedo et al, 2019; 2020; Keeley et al, 2020) ( Figure 6E , see Methods ). Specifically, we calculated the mapping weights to reconstruct each A1 neuron’s PSTH from those of MGBv neurons under different conditions.…”

Section: Resultsmentioning

confidence: 99%

“…In addition, as shown in Figure S6 F–G, we calculated the mapping vector from the Poisson generalized linear model (GLM) (Keeley et al, 2020). In the Poisson GLM, the instantaneous firing rate of A1 approximated by the above equation with a nonlinear kernel: …”

Section: Methodsmentioning

confidence: 99%

Computational Circuit Mechanisms Underlying Thalamic Control of Attention

Lam

Wimmer

et al. 2020

Preprint

View full text Add to dashboard Cite

SummaryThe thalamus is a key brain structure engaged in attentional functions, such as selectively amplifying task-relevant signals of one sensory modality while filtering distractors of another. To investigate computational mechanisms of attentional modulation, we developed a biophysically grounded thalamic reticular circuit model, comprising excitatory thalamocortical and inhibitory reticular neurons, which captures characteristic neurophysiological observations from the alert behaving animals. Top-down attentional control inputs onto reticular neurons effectively modulate thalamic gain and enhance downstream readout, to improve performance across detection, discrimination, and cross-modal task paradigms. Heterogeneity of thalamic responses plays an essential role in downstream decoding during attentional modulation. Dynamical systems analysis explains why reticular neurons are an especially potent site for top-down control, as implicated by experiments. Perturbation analysis reveals excitation-inhibition ratio as an effective parameter governing thalamic processing. These findings establish experimentally testable circuit mechanisms for attentional control in thalamus, with implications for distributed neural control of cognitive processing.

show abstract

“…Statistical relations of activity across brain regions have been traditionally used to infer causal interactions between them [4][5][6][7][8][9][10][11][12] . The majority of previous studies have relied on macroscopic measures of neural activity, such as the local field potential, EEG or fMRI BOLD signals [13][14][15][16][17][18][19] .…”

Section: Introductionmentioning

confidence: 99%

“…Likewise, in the macaque visual cortex, only a small subset of population activity patterns are statistically related to downstream activity 21 . Such findings have highlighted the importance of single cell resolution measurements in characterizing interareal interactions and have spurred the development of diverse multivariate statistical analysis methods to quantify the interactions between populations of neurons recorded across different brain areas [4][5][6] .…”

Section: Introductionmentioning

confidence: 99%

Dynamic causal communication channels between neocortical areas

Javadzadeh

2021

Preprint

View full text Add to dashboard Cite

Dynamic pathways of information flow between distributed brain regions underlie the diversity of behaviour. However, it remains unclear how neuronal activity in one area causally influences ongoing population activity in another, and how such interactions change over time. Here we introduce a causal approach to quantify cortical interactions by pairing simultaneous electrophysiological recordings with neural perturbations. We found that the influence visual cortical areas had on each other was surprisingly variable over time. Both feedforward and feedback pathways reliably affected different subpopulations of target neurons at different moments during processing of a visual stimulus, resulting in dynamically rotating communication dimensions between the two cortical areas. The influence of feedback on primary visual cortex (V1) became even more dynamic when visual stimuli were associated with a reward, impacting different subsets of V1 neurons within tens of milliseconds. This, in turn, controlled the geometry of V1 population activity in a behaviourally relevant manner. Thus, distributed neural populations interact through dynamically reorganizing and context- dependent communication channels to evaluate sensory information.

show abstract

Modeling statistical dependencies in multi-region spike train data

Cited by 38 publications

References 44 publications

Scalable Bayesian GPFA with automatic relevance determination and discrete noise models

Scalable Bayesian GPFA with automatic relevance determination and discrete noise models

Computational Circuit Mechanisms Underlying Thalamic Control of Attention

Dynamic causal communication channels between neocortical areas

Contact Info

Product

Resources

About