Tuesday 14 June 2016

#MechanismWeek (a week of posts commencing June 20th 2016)

Cognitive science is, in principle, the search to understand the mechanisms that cause our behaviour to look the way that it does. We run experiments designed to figure out the form of the behaviour to be explained, and we propose models that try to account for the behaviour. But how well are we doing, and can we do better?

It turns out that there is a rich and extremely useful philosophical literature about mechanisms. Specifically, there is a lot of clear and accessible work describing what mechanisms are, and, more importantly, how science can go about modelling those mechanisms. This literature has provided us with a wonderfully useful central focus for our ongoing work, and I wanted to walk through the key issues here. (We have covered this topic in a couple of posts - here and here - but there are several interlocking issues that I want to spell out one at a time).

Sabrina is in Warsaw June 23-25th attending the Mechanistic Integration and Unification in Cognitive Science conference, where she will present on how ecological information provides the key to mechanistic explanations in psychology.

To celebrate, there will be a new post from her every day of the week commencing June 20th 2016 on the topic of mechanisms, specifically cognitive mechanisms, and how to model them. Below is a list describing the upcoming posts and providing links to some useful reading.

We'd like to invite you all to play along, either in the comments or on Twitter (where we live as @PsychScientists; suggested hashtag #MechanismWeek). I've sketched out the week below, with recommended readings so yo can play along.

Mechanism Week posts:

Monday: Mechanisms and Models of Mechanisms
In this post, we will define what a mechanism is and the kinds of models of mechanisms you can have. We will discuss functional models but we will focus on dynamic causal mechanistic models as the gold standard model we should be aiming for, and talk about what such a model gives us.

Youtube: Talk by Bechtel on dynamic mechanistic explanation

Bechtel, W., & Abrahamsen, A. (2010). Dynamic mechanistic explanation: Computational modeling of circadian rhythms as an exemplar for cognitive science. Studies in History and Philosophy of Science Part A, 41(3), 321-333.

Cummins, R. (2000). How does it work?" versus" what are the laws?": Two conceptions of psychological explanation. Explanation and Cognition, 117-144.

Tuesday: Cognitive Models Are Not Mechanistic
Bechtel & Abrahamsen propose that cognitive science has a problem. Our research programme is not built in the way that enables the development of mechanistic models. Instead, we build various kinds of functional models that use capacities or components that don't neatly map onto real parts of the actual mechanism. Can these still explain things, though?


Bechtel, W. (2008). Mechanisms in cognitive psychology: What are the operations? Philosophy of science, 75(5), 983-994.

Weiskopf, D. A. (2011). Models and mechanisms in psychological explanation. Synthese, 183(3), 313-338.

Wednesday: Do Dynamic Models Explain?

One way to address the challenge posed by Bechtel & Abrahamsen is to deny it's a problem. While mechanistic explanations sound great, maybe we can do amazing science without them. This is the tack taken by Chemero and Silberstein, who argue that not every science can do the localisation and decomposition of real parts and processes required to get to mechanisms. They argue that cognitive systems are nonlinear and therefore nondecomposable and must be modelled as dynamical systems, and that this is perfectly fine.

Silberstein, M., & Chemero, A. (2013). Constraints on localization and decomposition as explanatory strategies in the biological sciences. Philosophy of Science, 80(5), 958-970.

Thursday: Ecological Mechanisms and Models of Mechanisms
Chemero and Silberstein make a mistake (we think). They go into the system at the wrong level, find no sensible parts, and instead of using that as evidence they started in the wrong place they abandon ship and say cognitive science can't do mechanisms. We think that there is a level where we can ground mechanistic models of cognition and behaviour. This level is ecological information,  and we illustrate our analysis using Bingham's model of coordinated rhythmic movement.


Golonka, S.,& Wilson, A. D. (2012). Gibson’s ecological approach - a model for the benefits of a theory driven psychology. Avant, 3(2), 40-53.

Blog posts:
Dynamic Mechanistic Explanations in Radical Embodied Cognitive Science

                  Thinking about representations in relation to mechanisms

Slides from a talk Andrew gave on ecological information and mechanisms

          Slides from Sabrina's talk in Warsaw

Friday: Cognitive Science Should Aim for Mechanistic Models
This post ties together what we've covered over the week, and argues that because we can develop mechanistic models when we ground them at the ecological level, we absolutely should start doing so. Functional and dynamical models can be useful, but because they do not refer to real parts their explanatory power is weak at best and they can easily lead to flawed predictions. We have always believed that psychology can do better than it is doing right now, and that path takes us towards dynamical causal ecological mechanistic models.

No comments:

Post a Comment