Project work

Project work

In the project work you will be select a topic related to the course, conduct an experimental work, write up the result in a short paper and present your results in a presentation. I have added approximate dates when i am expecting you to complete each step in the project work. Please share your work and data on the Google Sheets links and make it readable to everybody, and make it editable for the supervisor (Sverker Sikström). 

 

1. Work mates: Find students to work with. You can be 2 or 3 person in each group.

2. Decide on a topic:  The project work should be about natural and artificial cognition. Do not choice a topic that is too difficult for you to implement. Here are some alternatives

  • Comparing natural and artificial behavior (i.e., compare with LAB 3). Take an AI model and study the behavior of the model and compare it with behavior of humans.
    • For example: Take an text-generator AI algorithm and see if shows in and outgroup biases.
  • Analysing psychological data with AI (i.e., compare with LAB 2). Collect behavioral data and analyse them with some AI algorithm. 
    • For example: Conduct a psychological experiment with an open-ended question and a rating scale. Use NLP to quantify the text and machine learning to train it to the scale
    • For example. Find EEG or fMRI data and analyse them with machine learning
  • Simulate a behavioral phenomena (i.e., compare with LAB 1): 
    • For example: Create a neural network that simulates how word frequency influence recall and recognition performance in humans. 
  • Literature review. You may choose this option, but it is not recommended, as I prefer that you make an empirical study. If you do a review then it should be more extensive than an empirical study. Notice that you design instructitions below does not apply for a litterature reivew. Summarize the literature in a field including AI and psychology
    • For example: What is the literature how AI can be used for assessment of depression 

3. Supervision of design (on 10/11): Each group will have a 15 minut supervision of your choice of topic, hypothesis and design. This supervision is conducted one week into the project work. Sign up on the Course  Sheet. Make a one page summary of your idea including the following headlines: Title, background (a few sentences), hypothesis (max 2 sentences), proposed design (the rest of the page), questions (if any). 

4. Write a detailed design of your study, as well research question and hypothesis (done 17/11). Look at the methods section of three publications related your work. Write a method section that allows somebody that does not know anything about your project to EXACTLY replicate our study. The method section must include the following items; method, design, participants, material, procedure.

Here is a short example. 

This study compare evaluations of the emotions love and happiness narratives between humans and Chat-GPT3.

Research question. Are humans or Chat-GPT3 better at classifying emotions.

Hypothesis. Chat-GPT3 are better than humans in classifying narratives inte love or happiness.  

Method 

Design. The design is a 2 by 2, where factors are respondents (humans or Chat-GPT3) and the emotions (love or happiness).

Participants. 100 participants were recruited from Proflic.com, living in US that were English native speakers. Mean age 26.5 (sd 5.2), 62 women, 30 men, 8 preferred not to say. 

Material. We prompted to Chat-GPT to generate 10 narratives of love, and 10 stories of happiness, using the following prompt: "Write ten short stories describing the emotion ['love' or 'happiness']  

Procedure. The 10 narratives were shown to the participants, and they were asked to classify the narratives as either love or happiness. The following instruction/prompt were used: "Classify this story into either the emotions love or happiness: "Sam really liked to go out with Susanne".

Data processing. A chi2-test was used to test weather the number of correct classifications were larger for Chat-GPT3 than for humans.

5. Approval before data collection (on 22/11). Have your method section approved by the course leader. 

6. Collect your data (start 23/11): Do your study according to your design. 

7. Write a short paper (start 27/11): Look at three related paper. Try to write the paper with the same structure, details, etc as these papers. The paper should be between five and ten pages. Give the paper to another group for feedback. 

8. Feedback (11/12): Give feedback on another groups paper.

9. Respond to feedback (12-14/12). Change your paper according to the feedback that you receive from another group.

10. Presentation (15/12): Present your project work (see schedule). 

 

Expected time line:

Week 1 (45):  3. Supervision of design

Week 2 (46):  4. Write a detailed design of your study, as well research question and hypothesis.

Week 3 (47).  Wednesday: 5. Approval before data collection  6. Collect your data:

Week 4 (48).  7. Write a short paper

Week 5 (49).  7. Write a short paper

Week 7 (50).  Monday: 8 Feedback. Tuesday-Thursday: Respond to feedback. Friday: Presentation