Research

Emotions shape our conscious experience and guide our interactions with the world and people around us. Most of these emotions arise from external information that comes in through our senses 👁️👂👃👅✋.

The Computational Cognition & Affect Lab wants to know: How does perceptual information contribute to emotional experience?

We investigate the large-scale neural algorithms that the brain uses to process perceptual and emotional information. Specifically, we apply computational models of perception to human behavioral, self-report, and neuroimaging data collected while people experience emotional stimuli.

We use open-source scientific computing tools for our research. As members of the computational research community, we are also interested in meta-research about improving knowledge and use of scientific computing software.

PS: we sometimes apply tools of psychology and neuroscience to other personal interests!

Computational cognitive & affective neuroscience

Current projects include…

Perceptual building blocks for subjective emotions

Humans and other species share certain brain & cognitive systems for survival that are tuned to detect very specific perceptual inputs. (For example, detecting the expanding shape of something looming that’s on track to hit you in the face. Danger!) How does the information from these evolutionarily old systems contribute to human emotions in particular?

Relevant publications:

Components of mixed emotional experiences

Mixed emotions like relief or disappointment often arise from reversals of fortune, where something almost happens but doesn’t at the last minute. Can we identify the perceptual ingredients of mixed emotions, and differentiate them from emotions with similar perceptual inputs but different outcomes?

Meta-science of computational research methods

Current projects include…

Building community resources for scientists working on public datasets

Many different groups of scientists are often conducting simultaneous studies on the same public datasets. How can this scientific community share knowledge and expertise about the data, both independently from and in harmony with the original stewards of the data?

Relevant publications:

Contributing to open-source scientific software

We try to leave our software better than we found it! This often takes the form of collaborating with users and maintainers on GitHub to fix and improve functionality for packages that are used by many other scientists.

Sample contributions:

  • emonet-pytorch: PyTorch version of the EmoNet convolutional neural network for emotion classification from images (Kragel et al., 2019)
  • ggseg: R package for plotting whole-brain parcel statistics using ggplot2

Other projects

Other current projects include…

Memory mechanisms of trivia expertise

How are some people predisposed to remember lots of random facts?

Current research questions include:

  • How long after initial learning do trivia experts show enhanced episodic-semantic binding?
  • Can non-trivia experts be trained in expert-like memory strategies to improve incidental trivia retention?
  • Is trivia experts’ enhanced episodic-semantic binding associated with differences in temporal lobe structure and function?

Relevant publications:

This project is being conducted in collaboration with the Aly Lab at UC Berkeley.