One year ago I was hired as the head of the behavioral core (Neuro-BAU) at the Department of Fundamental Neurosciences (University of Lausanne, – Switzerland, @UNIL).
It’s been a year of success stories, learning new ways of doing research, and frustration, too.
The following is a reflection on the role of behavioral neuroscience in … well .. neuroscience, and the beginning of a course to promote a “cognitive & behavioral science culture” in an otherwise cellular/molecular-neuroscience academic environment.
“… someone in the group spotted an electrophysiology rig. The person commented with joy:<< Ah! So this where the real science is done!>>”
From time to time, I have the opportunity to show our behavioral core to visiting scholars and company representatives – they usually are really impressed by the platform. People like the idea of having a central resource that is also managed by someone who has deep knowledge in behavior.
However, sometimes I meet people that are not interested at all in the stuff I’m showing them. In general, that’s OK. However, sometimes this lack of interest is the reflection of a negative attitude towards behavioral science.
Here is an example:
Not long time ago I was showing the behavioral platform to a group of visiting scholars. As we moved from one wing of the building to another, we passed nearby a lab and someone in the group spotted an electrophysiology rig. The person commented with joy:<< Ah! so this where the real science is done!>>. The comment was not directly addressed to me but to another member of the group.
That was pretty rude – and I believe that this comment says a lot about the attitude towards behavioral science. An attitude that is driven, in my opinion, by a lack of knowledge about cognitive and behavioral science. Many people believe that behavior is about putting a rodent in a specific setting/arena and then look at what it does. But what should we look for?
Behavioral Neuroscience: Hunch-based Science?
Whenever I design a protocol in a certain way it is because I want to test a specific hypothesis and this hypothesis is derived from the knowledge of how certain cognitive processes are believed to drive the observable behavior.
However, I have frequently seen researchers (including myself!) fall for a bias quite common among practitioners (and “experts”, too) of behavioral neuroscience: the “introspection and hunch” bias.
“Introspection and Hunch” Bias
Introspection may go like this: << If I were in e I would do x>>
Where e is the experimental setting (this could be a watermaze pool, a conditioning chamber, an open arena with objects …) and x is some sort of loosely defined strategy (or heuristic) sampled from our own rich knowledge of folk psychology. This thought process distorts a proper, well thought, experimental design in favor of what we believe we would do in a similar solution (i.e. how we perceive the cues, how we explore the environment …).
In doing so, we fail to acknowledge that:
- We are not rodents (WTF?!).
- Most cognitive processes we are interested in may not be accessible to introspection.
- Introspection may only capture a distant reflection of a rational behavior we believe we are applying.
- There is a large corpus of literature that supports the choice of a particular protocol.
Often the process of selecting between two similar protocols is biased: a protocol described in a high-impact journal has more weight than a protocol from a less-know (but specialized) journal.
Behavioral experiments in high-impact journals (usually multidisciplinary-molecular-cellular neuroscience papers) are often treated as a “nice-to-have-but-not-critical”,”cherry-on-top-of-the-cake”,”just-because-reviewer-3-asked” experiments. As a consequence, the behavioral protocols (sometimes the behavioral data, too) tend to be poorly reported (?!?!).
Nonetheless, many people want to replicate the methods used in a high-impact journal (hoping this will yield a high-impact publication!?!), despite the lack of information on how to properly reproduce the behavioral experiment.
This “Introspection & Hunch” bias is quite common also in behavioral experts – myself included! However, years of training, experience in the field, and knowledge help us to look out for this bias.
Behavioral Neuroscience: Lack of Knowledge
People are more likely to fall for the “Introspection & Hunch” bias if they lack a basic knowledge of cognitive/behavioral processes.
During my PhD and post-doctoral experiences, I used to interact with people with a good (or exceedingly good!) knowledge of behavioral and cognitive science. So, whenever we had a conversation, there was always a basic shared understanding of the processes driving behavior and the factors affecting behavioral measurements. But this is not always the case when working with people from different fields.
From the conversations and discussions I had with scientists from an exclusively molecular / cellular background I have realized that we (me and my interlocutors) often do not have this comforting shared understanding of behavioral and cognitive processes. So, how do we go about to design a behavioral experiment if we don’t have this shared knowledge?
Teaching Behavioral and Cognitive Science Principles
I made my mission to create the shared understanding of cognitive & behavioral processes necessary to:
- have a productive interaction with me,
- develop together with me a behavioral protocol that is tailored to the specific needs of the researcher.
Having a simple input-output relationship (“I want a behavioral experiment that does x” –> “here is the experimental design you need”) won’t cut it.
I want to support researchers in the thought process that goes into designing a behavioral experiment aimed at testing a specific hypothesis.
So, I came up with this idea on how to introduce Behavioral and Cognitive concepts to scientists used to work, and think, at a different scale (e.g. molecular / cellular neuroscience).
I have designed a course titled: “Behavioral Neuroscience for Rational minds” (there is a hashtag for that #BN4RM) that is focused on suppressing the “introspection and hunch” bias by exposing the formalism of behavioral analysis and cognitive constructs.
The course is pretty exciting (and pretty dense, as well). It draws concepts from the tradition of artificial intelligence, passing by Lashley, Marr, Heider & Simmel, to build a framework (an actual and practical toolkit!) that can be used to develop and analyze behavioral experiments.
This course has the main goal to ease the transition from traditional behavioral assays to computational behavioral studies that take advantage of modern in-vivo imaging/recording techniques in freely-moving animals.
As a said, it is pretty dense, so it is the topic of my next post!