2.14: Brainy Programming with Chris Madan
You can find every episode of Bold Signals on Soundcloud, iTunes, and Stitcher. If you dig the podcast and have a few seconds to spare, kindly leave a review or star rating in any of those places. Season two episodes are also available for download via Figshare.
If you are interested in being a guest for an upcoming episode of Bold Signals Season 2, drop us an e-mail or check out this page. Thanks in advance!
This week we examine the relationship between emotion and memory with psychology and neuroscience researcher Christopher Madan. In our conversation we discuss everything from computational modeling and machine learning, to learning how to program in grad school, to navigating the uncertain terrain that is scientific jargon.
Want more info about Chris?
Chris on Twitter | Chris’s Website | Chris on ResearchGate
An Introduction to MATLAB for Behavioral Researchers
Want more info about Bold Signals and John?
Bold Signals on SoundCloud | Bold Signals on iTunes | Bold Signals on Twitter | Bold Signals on Facebook
John on About.Me
Links and Show Notes
The cover art for this episode is a compilation of images from the Pictures of Facial Affect (POFA) collection created by Paul Ekman and Wallace Friesman and used in their influential work categorizing facial expressions. This profile of psychology researcher Lisa Barrett includes a general (and nicely critical) review of Ekman’s work.
There is a lot still to be written about the leak of OKCupid user data under the guise of open science, but widely circulated blog posts by Emily Gorcenski, Oliver Keyes, and John Sakaluk have framed the initial conversation. For continuing coverage, watch this post by Micah Allen. Though the lack of de-identification makes this latest release especially egregious, this is not the first time OkCupid user data has been scraped. OkCupid co-founder Christian Rudder even used user data in his popular OkTrends blog and the resulting book Dataclysm.
Beliefs allegedly held by the individual at the center of the OkCupid controversy not withstanding, it would be disingenuous to discuss research ethics without mentioning the historical context of the Nuremberg Code. Together with the Declaration of Helsinki and the controversies surrounding experiments like those conducted by Stanley Milgram and Laud Humphreys, this code is the basis for regulations related to research involving human participants.
Recent developments in interrelated fields including genomics, citizen science, data sharing, and big data have profoundly complicated questions about what it means to consent and participate in a research study. The 2013 study of emotional contagion on Facebook is probably the most widely publicized case of a disconnect between the actions of a regulatory apparatus and the concerns of participants.
To hear more about Chris’s research on emotion and memory, listen to this episode of the excellent neuroscience-oriented podcast Neurocurious.
Chris currently works with Elizabeth Kensinger at Boston College. His other major collaborators include Jeremy Caplan, Esther Fujiwara, and Marcia Spetch at the University of Alberta, and Tobias Sommer at University Medical Center Hamburg-Eppendorf.
In modern psychology/neuroscience parlance, “behavioral” bears no relation to behaviorism- the approach that subsumed psychology for decades. Rather, it means something like “research not involving brain imaging or other biology-based measures”. First season guest Michael Greenstein’s experiments look superficially similar to Chris’s, but are considered “behavioral” because they don’t take place in an MRI scanner.
The relationship between emotion and memory is such a huge research area that it even has its own Wikipedia page. For an introduction to these topics from a cognitive neuroscience perspective, see this widely cited review.
The Center for the Study of Emotion and Attention at the University of Florida has developed and distributed a number of different stimuli sets usful for research involving emotion. Access requires special permission, but both the International Affective Picture System (IAPS) and Affective Norms for English Words (ANEW) sets are widely used throughout psychology and neuroscience.
For younger listeners, here’s what a 56k modem sounds like. To capture the full experience of using the internet in the era of dial-up, take the frustration of watching a YouTube video inexplicably fail to load past a certain point and apply it literally everything online.
The SPM suite, which is widely used to analyze brain imaging data, is probably the ur-example of neuroscientists creating and releasing useful pieces of software. Created by the Wellcome Department of Imaging Neuroscience at University College London, SPM is is made freely available to researchers in the brain imaging community. Similarly free software packages include AFNI, Freesurfer, and FSL. To find all manner of brain imaging-related software and resources, visit the Neuroimaging Informatics Tools and Resources Clearinghouse (NITRC).
As demonstrated by this paper by Anders Eklund and colleagues, the different assumptions underlying different brain imaging software packages can have startlingly major consequences when it comes to analyzing data.
When programming experiments, many psychologists and neuroscientists opt to control for order effects through counterbalancing rather than pure randomization.
Chris’s studies of word properties using computational modeling are described here and here. See this paper for a general overview of computational modeling in brain imaging research.
When considering the reality of machine learning, it’s useful to think about data analysis rather than the Terminator franchise. This Wired cover story and this “gentle introduction” provide nice entry points to machine learning, while a much more thorough overview can be found in this free Coursera course from Andrew Ng.
Multi-voxel pattern analysis (MVPA) is a technique for analyzing brain imaging data that involves programming an algorithm to search for patterns of brain activity that differentiate two or more experimental conditions. This paper provides a thorough overview of the principles underlying MVPA and related techniques.
Though Chris provides an excellent general definition, researchers working in neuroscience, psychology, and computer science conceptualize memory very differently. Compare these reviews of memory from the perspective of neuroscience, cognitive psychology, and cognitive neuroscience to the memory in your computer or mobile device.
Object recognition is a mind-numbingly boring, fundamentally fascinating, and endless rabbit hole of heated arguments. Even for researchers ostensibly working in the same field, the precise definition of “object” can vary greatly. Throw in the three pounds of complexity that is the human brain, and defining things gets exponentially more difficult.
Chris curated the Real Scientists Twitter account the week of January 17th to 23rd, 2016. Over the course of the week he led conversations ranging everywhere from the use of brain imaging to memory strategies. Look closely and you might even spot some input from a certain podcast host.
There is an ever increasing number of science-related rotational-curation Twitter accounts including Real Scientists, AstroTweeps, BioTweeps, NeuroTweeps, and I Am SciComm. In the name of interdisciplinary scholarship, also be sure to check out We the Humanities.
Especially in the case of fMRI research, there are A LOT of steps between the actual activity of cells in the brain and the signal analyzed by neuroscientists. Many fMRI researchers use “Regions [x] lit up.” as a shorthand for “After a series processing steps, comparisons of signals related to energy use reached statistical significance in a cluster of three dimensional pixels that, at least according to a standard atlas, correspond to the [x] region of the brain.”. Unless they’re using optogenetic techniques, most neuroscientists aren’t actually lighting up any brains.
Jody Culham’s site fMRI for Newbies (formerly fMRI for Dummies) is an excellent starting point for learning the terminology and methodology of fMRI research.
EurekAlert!, the service that underlies a substantial portion of science reporting, turned 20 years old the week of this episode’s release. Operated by the American Association for the Advancement of Science, EurekAlert! is essentially a giant clearinghouse for science-related press releases.
No seriously, most scientists on social media (including the host of this very podcast) are very happy to answer genuine questions related to their work or field. At the same time, scientists also tend to be extremely busy with their own research and homework. So please don't ask them to do yours for you.
All of Chris’s publications can be found on his website. His papers specifically related to mentorship and getting involved in research can be found here.
Chris’s book An Introduction to MATLAB for Behavioral Researchers is available for purchase via Amazon. If have access to a university or institutional library, you may also be able to read it for free through Sage Research Methods. Similarly, Matlab for Neuroscientists is also available on both Amazon and via ScienceDirect. There are lots of good ways to learn Python, including Codecademy, Google’s Python class, and the book Python Programming: An Introduction to Computer Science.
The German Academic Exchange Service (DAAD for Deutscher Akademischer Austauschdienst) is the largest German support organisation in the field of international academic co-operation.
Robert Epstein, research psychologist and author, recently wrote in Aeon that “Your brain does not process information, retrieve knowledge or store memories. In short: your brain is not a computer.” While the “brain as computer” metaphor is certainly problematic, it seems a bit excessive for a critique of taking the metaphor too literally itself takes the metaphor completely literally.
This week’s Bold Signals Book Club selection is Definitely Maybe by Arkady and Boris Strugatsky, a science fiction novel that takes place almost entirely in the minds and living rooms of scientists working in the Soviet Union. Both a dark comedy about the process of science and a uniquely claustrophobic take on dystopian science fiction, Definitely Maybe also seemingly predicted one of the most bizarre theories proposed about the Large Hadron Collider.
Subjected to Soviet censorship and poor translations for decades, the work of Arkady and Boris Strugatsky is slowly becoming widely available. Their two most acclaimed novels are probably Hard to be a God and Roadside Picnic, the later of which was the basis for Andrei Tarkovsky's classic film Stalker (which, along with Tarkovsky's other films, is currently available for free via the official Mosfilm Youtube channel).
If you’d like to purchase any of the books discussed in the book club and support the podcast at the same time, take a gander at our Amazon store. The podcast also has an Amazon wishlist if you’d like to donate a book. The full list of book club books is here.
Music in this Episode
“Enterprise 1” by Languis “Twilight Days” by Skipless “Time Doesn’t Exist, Clocks Exist″ by Mizontiq “XHD” by Aydio
DOI: 10.6084/m9.figshare.3394081