Read on Twitter
I'm rebooting this discussion on embodied cognition (as an outsider to the field) as a new thread. I've learned quite a lot reading this thread and thinking about the issues and thanks to @criticalneuro for pushing me to be clearer (1/n) https://twitter.com/bradpwyble/status/1317544072198881281
At issue is whether the strong embodied cognition(EC) perspective posits that physical interactions with the world (sensory and motor) form the basis of cognition writ large which we’ll call my “strawman” version of EC. disclaimer: I am not well read in EC (2/n)
I’ll try here to articulate my understanding, since Ida has asked for more detail. This is an important exercise in clarifying thoughts. Apologies in advance for what I get wrong and I’m looking forward to learning. (3/n)
I think it’s helpful to start with an familiar analogy since I’m a vision scientist: the idea that Gabors (or something vaguely like) serve as the basis for cortical vision (i.e. that they are the first level of representation within V1). (4/n)
The idea is that ~Gabors (a standin for receptive fields of V1 that are weakly understood and vaguely like oriented Gabors) are the first level of information representation in the cortical visual system, and later levels of processing inherit their information from them. (5/n)
This view can be an analogy for EC, which is that physical actions (and their consequent effects on sensation) are the basis set of cognition. That analogy seems reasonable but I don’t think it’s quite what strong EC says and here’s why. (6/n)
No one is claiming that ~Gabors should be the way that higher order visual function is construed (i.e. no one is a Gaborologist). In other words, ~Gabors are not considered the lens through which all cortical vision should be framed and discussed (7/n)
but I think this is what at least some of the stronger views in the EC field are claiming, i.e. that all cognitive functions should be viewed through the lens of physical interactions with the world. I'll provide some examples: (8/n)
Linda Smith in doi:10.1016/j.dr.2005.11.001
“The embodiment hypothesis is the idea that intelligence emerges in the interaction of an organism with an environment and as a result of sensory-motor activity” (9/n)
“The embodiment hypothesis is the idea that intelligence emerges in the interaction of an organism with an environment and as a result of sensory-motor activity” (9/n)
Andy Clark in DOI: 10.1016/j.tics.2006.06.012
“I suggest that words and sentences form part of the process of thinking, and that they do so not merely in virtue of their contents but also in virtue of their very materiality: ... (10/n)
“I suggest that words and sentences form part of the process of thinking, and that they do so not merely in virtue of their contents but also in virtue of their very materiality: ... (10/n)
their physical existence as encountered and perceptible items, as sounds in the air or as words on the printed page.”
“By materializing thought in words, we create structures that are themselves proper objects of perception, manipulation, and (further) thought.” (11/n)
“By materializing thought in words, we create structures that are themselves proper objects of perception, manipulation, and (further) thought.” (11/n)
My Interpretation: we use language only by projecting it to the sensory level. In other words, thinking through language in a fully amaterial sense does not occur, but rather we must collapse words to the sensory level in order to use them as a scaffold. (12/n)
In “Perceptual symbol systems”, Barsalou argues against representations that allow amodality (i.e. arbitrary symbolic representations). This includes many connectionist approaches in which internal layers act as a conceptual system. (13/n)
He suggests instead that we ground our thinking in simulations, including language
“A linguistic symbol is not an amodal symbol, nor does an amodal symbol ever develop in conjunction with it. Instead, a linguistic symbol develops just like a perceptual symbol.” (14/n)
“A linguistic symbol is not an amodal symbol, nor does an amodal symbol ever develop in conjunction with it. Instead, a linguistic symbol develops just like a perceptual symbol.” (14/n)
In contrast to these perspectives, I don't think a mainstream view in vision science is that we must project our higher-order representations down to the earlier levels to make use of them. (15/n)
In my brief review I wasn't able to find anyone who articulated the limits of this strong EC view, and if someone could point me to them I'd appreciate it. I also apologize for the brevity of this literature dive and any inaccuracies I've composed through it. (16/16)
