Course Syllabus
Psych 164: Brain Decoding
Instructor: Justin Gardner, PhD
Email: jlg@stanford.edu
Office Hours: If you have a question, then likely other students in class will have the same question, so bring it up in class. If your question cannot be answered this way, then you may sign-up for an office hours appointment here.
TA: Katherine Hermann
Email: hermannk@stanford.edu
Course Description: Can we know what someone is thinking by examining their brain activity? Using knowledge of the human visual system and techniques from machine learning, recent work has shown impressive ability to decode what people are looking at from their brain activity as measured with functional imaging. The course will use a combination of lectures, primary literature readings, discussion and hands-on tutorials to understand this emerging technology from basic knowledge of the perceptual (primarily visual) and other cognitive systems (such as working memory) to tools and techniques used to decode brain activity.
Meeting time: TuTh 10:30-11:50 Spring quarter
Location: Online (Zoom links are available from the zoom tab on the left)
Pre-requisite: Psych 30 Introduction to Perception or Psych 50 Introduction to Cognitive Neuroscience or Consent of Instructor
Course objectives: The overall objective is to learn how brain decoding works! We aim for an understanding that will let you assess what is possible with current technology and what limitations there may be. The purpose of this course is not as much engineering as it is understanding. By the end of the quarter you should be able to determine what is actual state-of-the art brain decoding from what is fantasy. If you see an article in the popular press about some new miraculous brain-decoding result, you should be able to apply knowledge learned in the course to understand what was actually done and tell for yourself whether you think the technology lives up to the hype.
We will also stress the ability to read, understand and communicate methods, findings and interpretation of cutting-edge research. What this means, is that by the end of the course, you should be able to identify a paper about brain decoding you are interesting in reading, read it, figure out what was done, how it was done and what it means (or doesn’t mean) and then be able to explain all of that to a friend (who is smart and cares about brain decoding!) who hasn’t read the paper. As many primary research articles have mathematical and technical details that you may not already have experience or background understanding, another course objective is to build your skills at reading around obscure quantitative descriptions to being able to understand and communicate core principles.
Writing is an important way in which you can demonstrate understanding and communicating that understanding to others. So this course will also develop your ability to communicate complex ideas in written from through weekly writing assignments that include revisions. This course fulfills the Writing in the Major (WIM) requirements for psychology majors.
Course topics: We will set out through the recent literature on brain decoding. We will focus primarily on the decoding of visual perceptual experience because work in this area is quite advanced and much of the pioneering work has been done in visual decoding. Also, by focusing on visual perceptual decoding there will be continuity from paper to paper so that you can build on the knowledge you gain throughout the quarter. At the end of the class, you will be able to steer the discussion to other areas that you are interested in by choosing a paper related to your interest and running the discussion as a group.
Course structure: Each week we will tackle a single primary literature paper related to brain decoding. On Tuesday there will be a lecture that sets the stage for understanding that paper, giving you necessary background and an overview of technical and conceptual issues that is intended to assist you in reading the paper.
Everyone is expected to read each paper each week, as thoroughly as possible. What does thoroughly mean? These papers can be complex. Details can be hidden in the methodology. Strange terminology may be used. Expect to read the paper a few times before coming to understand it. If you need to go check what a word means on wikipedia so that you can understand a paragraph, do that! If a paper is referenced that contains information that might help you, look it up and take a look. Yes, this can take a few hours or more to get through a single paper, but that’s normal. Even experienced researchers in the field like myself will take many hours reading a paper and looking up background information.
A short written synopsis is due by midnight on Wednesday. The synopsis should include either a question, a comment or a thought about the paper that can be discussed in the group discussion. We will then have a guided discussion of the paper stressing the content of the paper and how to understand and interpret the data presented. The goals of the discussions are two-fold: understanding issues at the forefront of brain decoding and developing skills to critically read, interpret and communicate the content of papers in general. The following Monday you will hand in a brief self-critique of your synopsis. The self-critique should take the form of an edited version of the synopsis you wrote where you clarify your writing and highlight what you have learned from the discussion. You may also add any comments you have on the discussion itself.
To practice your ability to critically read and communicate primary literature, you will work as small groups (2-4) to choose a paper to read and present to the class. With your designated group you will choose a paper related to brain decoding. The paper may be taken from things about decoding that are currently in the news - like mind reading, or lie-detection. After finding a suitable primary literature paper, you will make a short presentation (up to 30 minutes) to set the stage for the paper and then run a group discussion for the rest of the class.
Readings: Will be available on Canvas. You may also download directly from the internet, I recommend looking for papers in pubmed https://www.ncbi.nlm.nih.gov/pubmed/ rather than google as pubmed is a curated database that has proper links to primary source material. If you insist on google, then make sure you are using google scholar: https://scholar.google.com
Paper synopsis
Each week I would like you to write a short summary of the paper that you have read. This does not have to be an extensive essay, a paragraph or two for the following five questions (please make sure to use your own language and not copy language from the paper. Trying to explain with your own words what the authors did is critical for understanding the paper!)
- What is the question that the authors are trying to address? Try to state this in plain english what the authors wish to achieve.
- What do they do to address this question? If experimental, what are the experiments that were done? If modeling, what is it that they modeled?
- What did they find? This should be as literal as possible - try to separate what the experimenters conclude from what data are presented.
- What did they conclude given what they found? Here try to state what the conclusions of the paper are based on the results.
- A question you have. The question may take the form of a criticism of what is concluded, a detail of the methodology that may change the way the results are interpreted, a possible alternative explanation for the results, an idea for a new experiment based on the paper or simply a question about something you did not understand very well and wish to understand better. This question (or questions) should be a possible discussion point for the Thursday meeting.
Synopsis self-critique
The goal of group paper discussions is for everyone to get a better understanding of the papers we read by learning from each other. Did the discussion work? Go back and read your synopsis from before discussion and decide if your understanding of the paper has changed. If so, what has changed it? If you understand something better that you did not understand before, what did you understand? If you have a new question about how to interpret the research or an idea for a new experiment that can be done, write that down. Alternatively if there is something that still is not clear to you about the paper that you wish to answer then feel free to write that as well.
Please also feel free to comment on the discussion environment of the class itself. Class should be a safe place to discuss and learn form each other. If you feel like it is not serving this purpose of if you feel there is something that could be improved, let me know!
Group presentations
You will work in groups to select and present a paper about brain decoding. Each group will present a single paper with some slides to set the background and then run a discussion of the paper. Each member of the group should have a clearly defined role in the class presentation (i.e. one member presents background, another the methods, another the results...). The goal is to really dig into the paper you have chosen and help others in the class do the same. So, please hit the major points, but also make sure to be very clear about what was done and how it was done. Note that you don't have to agree with everything in the paper - if you have a criticism, just be very clear about the critique and how it affects understanding of what the paper has achieved. For some possible papers to present, check: Suggestions for presentation papers
Grading basis:
30% Paper synopsis. Due each week by midnight on Wednesday
20% In-class participation. Each Thursday we will have a group discussion of a paper for which we will expect all participants to have read thoroughly the paper so that we can discuss and understand it together. This is a major part of the class. Your grade will be based on attendance and participation.
10% Synopsis self-critique. Due the following week by midnight on Monday after a paper reading.
20% Group presentations. In small groups (2-4), choose a paper related to either mind reading or lie detection and make a background presentation and run a discussion.
20% Final paper. Given what you have learned across the quarter, you will write a primer on brain decoding. The piece should be written with the voice of an expert and provide evidence for your argument from papers in the literature. Length should be 1000 - 1500 words. You will work on this paper over the quarter, beginning with a preliminary discussion at the beginning of the quarter. You will later write a draft version of this piece which will be peer reviewed in one of the final sessions of the class. Finally, you will edit this draft using the comments you receive for the final version. Check the Final Paper - First Draft assignment for more details.
|
Date |
Week |
Topic |
Paper(s) |
|
Tu: 3/30 |
1 |
Discussion of a classic paper by Hubel and Wiesel about how the first stage of visual processing in the cortex encodes visual stimuli. This basic understanding of visual properties of neurons in the cortex will form the foundation of numerous subsequent readings in which visual responses are decoded from brain activity.
The background reading is not required, but are fun reads! |
Background reading (not required): Lettvin, J. Y., Maturana, H. R., McCulloch, W. S., & Pitts, W. H. (1968). What the frog’s eye tells the frog’s brain. In W. C. Corning & M. Balaban, The Mind: Biological Approaches to its Functions (pp. 233–258). This is just a cool little paper with a cool title. It proposes an ecological approach to understanding visual function in the brain. That is they try to understand why neuron's in the frog brain behave the way they do, by trying to understand what the frog needs vision for. The first paragraph or two is truly a classic way to start a paper.
There are also two chapters from an easy and fun to read textbook on vision for those of you with less background. Basic Vision, Chapter 1 & Chapter 2.
|
|
Th: 4/1 |
|||
|
Tu: 4/6 |
2 |
Decoding of the orientation of visual stimuli from human visual responses.
Background reading is an overview of the paper and might be better to read first. |
Kamitani Y, Tong F (2005) Decoding the visual and subjective contents of the human brain. Nature Neuroscience 8:679–685. http://doi.org/10.1038/nn1444 Background reading (required): Boynton, G. M. (2005). Imaging orientation selectivity: decoding conscious perception in V1., 8(5), 541–542. http://doi.org/10.1038/nn0505-541 |
|
Th: 4/8 |
|||
|
Tu: 4/13 |
3 |
Can working memory representations be decoded from visual cortex? |
Harrison, S. A., & Tong, F. (2009). Decoding reveals the contents of visual working memory in early visual areas. Nature, 458(7238), 632–635. http://doi.org/10.1038/nature07832
|
|
Th: 4/15 |
|||
|
Tu: 4/20 |
4 |
But, why does orientation decoding work with fMRI? Is it decoding columns or maps? |
Freeman, J., Brouwer, G. J., Heeger, D. J., & Merriam, E. P. (2011). Orientation Decoding Depends on Maps, Not Columns. The Journal of Neuroscience, 31(13), 4792–4804. http://doi.org/10.1523/JNEUROSCI.5160-10.2011 |
|
Th: 4/22 |
|||
|
Tu: 4/27 |
5 |
Kay, K. N., Naselaris, T., Prenger, R. J., & Gallant, J. L. (2008). Identifying natural images from human brain activity. Nature, 452(7185), 352–355. http://doi.org/10.1038/nature06713 |
|
|
Th: 4/29 |
|||
|
Tu: 5/4 |
6 |
Huth AG, de Heer WA, Griffiths TL, Theunissen FE, Gallant JL (2016) Natural speech reveals the semantic maps that tile human cerebral cortex. Nature 532:453–458. https://www.nature.com/nature/journal/v532/n7600/pdf/nature17637.pdf |
|
|
Th: 5/6 |
|||
|
Tu: 5/11 |
7 |
|
|
|
Th: 5/13 |
|||
|
Tu: 5/18 |
8 |
|
|
|
Th: 5/20 |
|
||
|
Tu: 5/25 |
9 |
Final assignment (brain decoding primer) peer-review session |
|
|
Th: 5/27 |
|
||
|
Tu: 6/1 |
10 |
|
|
|
Th: 6/3 |
|
Class attendance and participation
The learning objectives of the course will be achieved primarily through attendance and participation. Please be on time and come to class having read the paper and try to participate.
In understanding complex papers it is very useful to try to go through each part of the paper in as detailed a fashion as possible. We will often talk about seemingly trivial details of the paper, because sometimes the most trivial details are the most important ones. Often you will find that explaining the details of how an experiment was done or how it was analyzed helps you gain clarity and understand it better.
While group discussions always take some time before everyone feels comfortable they are not meant to be intimidating. There are no bad questions, and everyone will have some things they don’t understand or misunderstand. If you are confused about something there will be others who will be as well.
Late Work policy
Preparing a summary each week is intended to help you organize your thoughts about the paper you have read and will be critical as a starting point for each discussion. For this to be useful, you will need to complete it by Wednesday at midnight. The self-critique of the synopsis is intended to solidify what you have learnt from group discussion. Articulating and writing down major points will help you solidify what you have learnt. This should be done as soon as you can after class, but by the Monday after discussion at midnight at the latest.
You may drop at most 2 synopses (and the corresponding self-critique).
Missed class policy
This will be a small class that meets weekly and discussions are the most important part of the course. Please notify in advance (by email to jlg@stanford.edu) if you need to miss a class due to a conflict or illness.
Students with Documented Disabilities
Students who may need an academic accommodation based on the impact of a disability must initiate the request with the Office of Accessible Education (OAE). Professional staff will evaluate the request with required documentation, recommend reasonable accommodations, and prepare an Accommodation Letter for faculty dated in the current quarter in which the request is being made. Students should contact the OAE as soon as possible since timely notice is needed to coordinate accommodations. The OAE is located at 563 Salvatierra Walk (phone: 723-1066, URL: http://oae.stanford.edu).
Honor code
We will abide by the Stanford University Honor code. Please familiarize yourself with it: http://studentaffairs.stanford.edu/communitystandards/integrity/honorcode
Course Summary:
| Date | Details | Due |
|---|---|---|