This site uses cookies

By using this site, you consent to our use of cookies. You can view our terms and conditions for more information.

Gaussian process joint models for estimating latent dynamics of brain and behavior

Giwon Bahg
Vanderbilt University ~ Psychology
Daniel G. Evans
The Ohio State University, United States of America
Mr. Matthew Galdo
The Ohio State University ~ Psychology
Dr. Brandon Turner
The Ohio State University ~ Psychology
Chris Donkin
LMU Munich ~ Psychology
Prof. Andrew Heathcote
Univeristy of Amsterdam ~ Psychology

For bridging implementational and algorithmic levels of analysis (Marr, 1982), Bayesian joint models (Turner et al., 2013) have been applied for investigating shared statistical constraints between neural activities and cognitive model parameters. However, the previous joint modeling approach has assumed linearity in its linking function, which might not always be ideal when dealing with complex brain dynamics. Moreover, joint models based on covariance estimation often sacrifice the temporal dynamics of cognitive processes. As a solution to these limitations, we propose a Gaussian process joint model (GPJM), a data-driven and nonparametric joint modeling framework based on hierarchical Gaussian process latent variable models (Lawrence & Moore, 2007). In the GPJM, latent Gaussian processes serve as a linking function and model temporal dynamics governing neural and behavioral observations. The GPJM can incorporate spatiotemporal covariance structures as its constraints and evaluate the relevance of each latent dimension to the process of data generation. To verify the utility of the GPJM, we tested the model performance with simulation and an application to real data. The simulation results showed that the GPJM estimates cognitive dynamics while exploiting spatiotemporal constraints. In an fMRI experiment based on a continuous motion-tracking task, the GPJM explained the neural and behavioral data appropriately and also revealed non-trivial underlying dynamics that generate the data. Cross-validation analyses demonstrated that the latent dynamics trained with complete neural data and partially observed behavioral data could predict test data reasonably. How the latent dynamics could be interpreted is an open question.



Gaussian process
joint modeling
cognitive dynamics
spatiotemporal modeling


Cognitive Modeling
Models of Physiological Data
normality assumption Last updated 3 years ago

I echo the previous commenters' congratulations with your PNAS paper. Well done! I was wondering (as a person somewhat naive to the intricacies of joint modeling) whether the assumption of a multivariate normal distribution is reasonable given that a lot of our behaviour is governed more by ex-Gaussian processes (and I am not sure whether neural da...

Dr. Marieke Van Vugt 1 comment
Out-of-sample prediction of behavior Last updated 3 years ago

Thank you for your talk! I am wondering how well this joint neural-behavioral model would perform in cross-validation compared to cognitive models of the behavioral data. This is a general question as I'm not sure how to model continuous joystick data with a cognitive model, perhaps some continuous sequential sampling models with multiple cogni...

Dr. Michael D. Nunez 2 comments
Great talk Last updated 3 years ago

Great talk, very clear with beautiful slides! GP seems the logical next step to model the link function. Obviously, you may also use GPs to model other components of your joint modeling framework, such as the neural/behavioral response models. CONGRATS on your PNAS paper.

Prof. Jay I. Myung 1 comment
Cite this as:

Bahg, G., Evans, D., Galdo, M., Turner, B., Donkin, C., & Heathcote, A. (2020, July). Gaussian process joint models for estimating latent dynamics of brain and behavior. Paper presented at Virtual MathPsych/ICCM 2020. Via