Sonia Bishop, Professor

Closed (1) (1) Biased priors or inference in depressed and anxious individuals?

Closed. This professor is continuing with Fall 2018 apprentices on this project; no new apprentices needed for Spring 2019.

General to all projects listed:
Computational models have been a powerful tool for studying decision-making in both psychology and neuroscience. They have recently become popular in psychiatry as well. Part of the appeal has been that computational approaches delineate individual differences in decision-making that can explain why people with different psychiatric disorders (such as anxiety or depression) often make poor decisions.
Our lab investigate potential decision-making biases exhibited by anxious and depressed individuals. To do so, we leverage behavioral experiments, fMRI, and computational models inspired by Bayesian statistics and reinforcement learning algorithms from AI.
Specific to this project:
This study aims to investigate how anxious and depressed individuals update beliefs about themselves. Specifically, we will look at whether anxious and/or depressive disposition (trait anhedonia, trait anxiety etc.) are associated with (1) excessively negative prior beliefs, (2) asymmetric updating of these beliefs for positive or negative feedback, and/or (3) different updating for beliefs about the self vs. others. Teasing apart the contribution of these potential biases will require the use of computational models.


RA primary responsibilities will include data collection and assisting in data analysis. They are also expected to attend and participate in lab meetings.
Behavioral data collection will consist of administering online sessions for RPP participants, as well as in-person sessions in our lab. Code for the experiment has been written, but edits to the experimental code may need to be made during piloting. The experimental code is written in javascript and python and runs on a website hosted on Amazon’s web services.
Data analysis will consist of downloading data from web servers, organizing and cleaning data, and doing basic descriptive statistics and plotting. The RA may also help in fitting computational modes or doing other complex statistical analyses. Analyses will be done in python (and sometimes R).


Day-to-day supervisor for this project: Christopher Gagne, Graduate Student

Qualifications: RA must have some programming ability (preferably in python). Experience running participants in psychology experiments is also useful.

Weekly Hours: 9-11 hrs

Related website: http://bishoplab.berkeley.edu

Closed (2) Designing an app to tracking influences on anxiety and depression on daily decision making

Closed. This professor is continuing with Fall 2018 apprentices on this project; no new apprentices needed for Spring 2019.

General to all projects:
Computational models have been a powerful tool for studying decision-making in both psychology and neuroscience. They have recently become popular in psychiatry as well. Part of the appeal has been that computational approaches delineate individual differences in decision-making that can explain why people with different psychiatric disorders (such as anxiety or depression) often make poor decisions.
Our lab investigate potential decision-making biases exhibited by anxious and depressed individuals. To do so, we leverage behavioral experiments, fMRI, and computational models inspired by Bayesian statistics and reinforcement learning algorithms from AI.
Specific to this project:
We have previously shown that anxiety is linked to difficulties adjusting behavior to the volatility of the environment (i.e. whether one's actions tend to result in similar or different outcomes across time) - Browning et al. Nature Neuroscience,2015. We are interested in whether this also holds in a real-world setting. To study this, we need to program an app into which volunteers can enter their daily choices and the outcomes of those choices. We also seek to create an app-based version of our existing task.

To assist in developing of these two apps. Learning outcome: understanding the pros and cons of . naturalistic real-world vs controlled laboratory experiments and to learn how to translate from one to the other. If a second semester if undertaken, to collect data and learn how to analyze it using Bayesian statistics. To better understand the forefront of the field of computational psychiatry.

Day-to-day supervisor for this project: Sonia Bishop, Graduate Student

Qualifications: CS major or minor; strong programing ability ; in particular experience with development of apps for mobile devices

Weekly Hours: 9-11 hrs

Related website: http://bishoplab.berkeley.edu/

Closed (3) Effort based decision-making

Applications for Spring 2019 are now closed for this project.

General to all projects:
Computational models have been a powerful tool for studying decision-making in both psychology and neuroscience. They have recently become popular in psychiatry as well. Part of the appeal has been that computational approaches delineate individual differences in decision-making that can explain why people with different psychiatric disorders (such as anxiety or depression) often make poor decisions.
Our lab investigate potential decision-making biases exhibited by anxious and depressed individuals. To do so, we leverage behavioral experiments, fMRI, and computational models inspired by Bayesian statistics and reinforcement learning algorithms from AI.

Specific to this one:
This study will investigate how individuals differ in the valuation of effort during decision making. Willingness to expend effort in order to obtain rewards and avoid harm may be a crucial distinguishing feature among individuals vulnerable to anxiety or depression.


Day-to-day supervisor for this project: Jennifer Senta, Graduate Student

Qualifications: RA responsibilities: For this project, the RA will help in the process of designing and collecting data for a new study. Study design will be driven by Dr. Bishop and Jennifer Senta and Chris Gagne (graduate students), coding of the experiment will be a joint effort between Jennifer, Chris and the RA, as will data collection and data analysis. Data collection will be conducted in-lab in Tolman hall. RA’s are also expected to attend and participate in lab meetings. Qualifications: RA must have some programming ability (preferably in python). Experience running participants in psychology experiments is also useful.

Weekly Hours: 9-11 hrs

Related website: http://bishoplab.berkeley.edu/