Michael Silver, Professor

Related website: http://argentum.ucbso.berkeley.edu

Closed (3) Mapping ambiguous images in the brain

Applications for fall 2021 are now closed for this project.

Ambiguous images are scenes that contain more than one category (i.e., both an aquarium and a restaurant). We are interested in studying the ways in which different systems (computers, behavior, and the brain) prioritize a particular category when multiple types of information are present in an a given image. Understanding how ambiguous images are classified can help us explore the similarities and differences between artificial systems and humans as well as expand our understanding of where and how ambiguity is represented in the brain.

The study will ultimately consist of three parts:

(1) Using pre-trained neural networks to build a dataset of ambiguous images
(2) Measuring behavioral responses to ambiguous images
(3) Predicting cortical responses to ambiguous images using fMRI


We are currently working on step (1): building a dataset. Over the course of the fall semester, we will begin running subjects in step (2): behavioral responses. It is less likely that there will be time to officially begin step (3): fMRI responses this semester, but there will be opportunities for interested students to learn more about neuroimaging.

Your tasks would include:
- Image processing/analysis using Python
- Helping to design and run behavioral experiments
- Recruiting subjects

Day-to-day supervisor for this project: Julie Self, Graduate Student

Qualifications: Desired Skills: - Some exposure to introductory neuroscience concepts - Basic coding abilities in Python In the Silver lab, we are committed to making research accessible to everyone, and we strongly encourage applicants from underrepresented groups in STEM to apply.

Weekly Hours: to be negotiated