Machine learning approaches in Cognitive Neuroscience
Welcome to our module on machine learning approaches in cognitive neuroscience!
Cognitive Neuroscience is an interdisciplinary field investigating the neural basis of cognitive processes such as perception, memory, language, decision-making, and emotions. Our cognition, emotions and behaviors are a product of our brains and cognitive neuroscientists aim at creating neurobiological models of the mind to answer to questions like: How does the brain process information? What neural pathways are involved in memory formation? How do different regions of the brain communicate during complex tasks?
For that, scientists look at measures of brain activity and correlate it with behavior. Several methods are available to measure brain activity. Besides other things, the different methods differ in their spatial and temporal resolution. Two of the most used methods are fMRI (functional magnetic resonance image) and EEG (electroencephalogram).
Traditionally, the brain data collected with these techniques is analyzed with univariate approaches. That is, statistical methods that explored neural activity by contrasting conditions in pre-specified brain regions and/or processing times. The development of multivariate statistics, such as the Multivariate Pattern Analysis (MVPA), allows scientists to decode contents of perception and thought from patterns of neural activity, which empowers researchers with the ability to answer questions that traditional techniques could not.
Moreover, in the past years, the development of Brain-computer interfaces (BCIs) has offered a unique opportunity for humans to steer machines with the contents of their minds.
In the two weeks ahead, we will introduce you to the fusion between MVPA and Cognitive Neuroscience!
We will have two lectures, followed by two seminars on the topic. See info below:
Lecture 1 - Mikael Johansson will introduce you to Cognitive Neuroscience and its methods. He will present the challenges with traditional univariate approaches to analyzing brain data and the benefits of more recent multivariate approaches. Finally, Mikael will show and discuss BCI applications.
Lecture 2 - Inês Bramão will show you applications of the MVPA approaches to investigate the temporal dynamics of learning and memory. The lecture will start with a brief introduction to memory. After and through the discussion of illustrative examples, we will discuss how MVPA has been used to elucidate the temporal dynamics of memory function.
PSYE60_CogNeuroApplications.pdf Download PSYE60_CogNeuroApplications.pdf
Seminar 1 - Journal club on MVPA in cognitive neuroscience research
In this seminar, you will work in groups and present an article that used MVPA to investigate the neural basis of cognition and emotion.
Group 1, 2, 3 and 4 are to present between 13:15-15:00.
Group 5, 6, 7 and 8 are to present between 15:15-17:00.
You are free to select the articles yourselves.
Be sure that: 1) the article has used MVPA as a method to evaluate human brain data; 2) the article is recent (> last 5/6 years); 3) it is published in a peer-reviewed journal.
Seminar 2 - Journal club on "Mind reading" in Brain-computer interfaces
In this seminar, you will work in groups and present an article about a BCI implementation.
Group 1, 2, 3 and 4 are to present between 8:15-10:00.
Group 5, 6, 7 and 8 are to present between 10:15-12:00.
You are free to select the articles yourselves. Be sure that: 1) the article tests a possible BCI implementation; 2) the article is recent (> last 5/6 years); 3) it is published in a peer-reviewed journal.
For the presentations on both seminars
-> Please check your group here: https://canvas.education.lu.se/courses/23418/groups#tab-22850 . The groups are the same for both seminars.
-> Prepare a power-point presentation and be ready to give a presentation of about 10/15 minutes. You do not need to bring up all the points of the article. Focus on the aim of the article, describe the method in general (make sure that the tasks/conditions used are clear), and describe the main results and conclusions. End your presentation by including, at least, two questions to be discussed by the whole group (the questions can be directly tied to the article or more general).
-> In the end of the session, the group leader (who was randomly selected) is expected to send us the presentation (ines.bramao@psy.lu.se; mikael.johansson@psy.lu.se). Make sure that the names of everyone involved in the work are visible in the presentation.
Obligatory readings
Bramão, I., & Johansson, M. (2018). Neural Pattern Classification Tracks Transfer-Appropriate Processing in Episodic Memory. eNeuro, 5(4), ENEURO.0251-18.2018. https://doi.org/10.1523/ENEURO.0251-18.2018 
PDF
Norman, K.A., & Polyn, S.M., Detre, G.J., & Haxby, J. (2006). Beyond Mind-Reading: Pattern Analysis of FMRI data. Trends in Cognitive Sciences 10(9), 424-430. PDF Download PDF
King, J.-R. & Dehaene, S. (2014). Characterizing the dynamics of mental representations: the temporal generalization method. Trends in Cognitive Sciences 18(4), 203-210. PDF Download PDF