What is the meaning of representations?
This week there will be no Life and Mind seminar. Instead, I thought that we could try a new format which focuses on the blog, and thus make our discussion more available to the wider online community.
To start things off I have chosen a controversial topic in cognitive science that has been discussed in the context of the Life and Mind seminars before, namely representations (again!), and I’ve prepared a short position statement (2 pages) that is intended to provoke some comments. The main ideas of this ‘open letter’ do not relate to any technical details of enactive cognitive science, so I hope that they will be accessible to everyone.
To begin with, I would like to draw attention to two fundamental kinds of discourse in which the notion of representation is used in the cognitive sciences. I will refer to them as ontological and epistemological. Whereas the former concerns the “being” of a representation (i.e. what it is, where it is located, what is its material basis, etc.), the latter refers to the notion’s place in our system of scientific practice and knowledge (i.e. what phenomena does it explain, in what kind of theoretical framework, what are its premises, etc.).
In the ontological discourse the notion of representation is sometimes not used as an explanatory concept, but rather as a symbolic label for some observed phenomenon, such as a neural ‘emulator’ circuit. For me this kind of usage is okay, since it is possible that we can agree that such a circuit can indeed be distinguished in the nervous system. I would not choose to refer to such systems in this manner, but let’s not argue over labels.
Another common example of ontological usage is when the theoretical notion of representation is simply reified into some externally existing structures in the head (or organism, or even organism-world). In these cases the existence of certain representations inside the cognitive agent is simply presupposed, and the particular phenomenon the notion is supposed to pick out is left ambiguous.
I will not spend too much time discussing this kind of ontological usage here. For me, it just appears to be the usual outcome of standard scientific practice, whereby theoretical constructs start to become posited as real, external entities whenever they have become indispensable for the working life of a particular normal scientific community (cf. Latour). I only wonder whether this process of reification due to scientific practice could be analogous to the process of externalization found in skilful users of sensory augmentation devices, such as Bach-y-Rita’s TVSS or the Enactive Torch?
In any case, as philosophers we can resist this temptation of reification, and thus bracket the ontological questions in order to investigate the origins of the theoretical construct within the scientific discourse itself. On this view, the more pressing question is whether the notion of representation is a useful scientific concept that can, with certain assumptions, have explanatory value. Accordingly, for the rest of this post I will treat this notion as a concept that might help us to explain certain phenomena that are of cognitive scientific interest.
But even if we re-interpret the notion of representation as a useful regulatory concept that can help us to explain mind and cognition, we can still question this epistemological role. More precisely, I suggest that identifying it as a theoretic construct opens up the possibility of an account of its historical genesis. We need to understand its role in the wider context of scientific practice in which it is embedded in order to better evaluate its status.
Thus, coming from an enactive perspective, I propose to approach the mainstream cognitive science community almost like an anthropologist studying another culture. What does the notion of representation mean to those scientists and philosophers? Why is it so important to them, though we, in the enactive community, seem to have no use for it? In order to answer these questions we have to understand the background practices, the underlying goals and motivations, and their wider social context. Only then can we begin to understand why the notion of representation is one of the most valued elements of that approach such that almost nothing can throw it into doubt. And only then can we begin to communicate about how to properly move this debate forward.
From what I have read and heard from the cognitive science mainstream, I suggest that there are two crucial underlying motivations that make the notion of representation its central and indispensable element:
First, I noticed that the notion of representation is closely associated with a traditional mechanistic (materialistic?) view of the mind. Here it appears to essentially fulfil the role of protecting what makes us human from the clutches of relentless mechanical reduction. Thus, one of the main explanatory roles of representations is to act as a safeguard of our autonomy, usually in terms of making environmental detachability possible. In this way it is supposed to explain (ensure?) that we are not completely determined by the immediate demands of our surroundings.
Nevertheless, I think that we can diagnose this common usage as a symptom of lacking clarity about what makes us autonomous agents. Since the notion of representation must eventually be cashed out in some mechanistic manner, it ultimately does not buy us the kind of intrinsic openness it attempts to provide. What is needed to account for this openness is an organization that is inherently open, as is the case, for example, for a system whose operations determine its own systemic identity (expressed more poetically: an agent whose being is its own doing).
Second, it is sometimes argued that we need representations because otherwise the mechanisms do not “look like” what is going on during reflection and human cognition. For example, since we experience “aboutness” (e.g. a belief, an intention, etc.) it is argued that there must be something in our mechanistic realization as such a system that is about that something else, too.
I think that we can diagnose this stance as another symptom, namely of lacking a clear personal/sub-personal distinction. There is no reason to assume that there will be such a naïve isomorphism between these two distinct domains of discourse. To posit personal-level-like phenomena at the sub-personal level trivializes (and distorts) their relation in the form of simple duplication. It is a re-stating of the phenomenon to be explained, rather than a form of explaining it. As such, it does not bring us any closer to understanding cognitive agency.
These rejections of the notion of representation might be rather quick and dirty, but the main point for me is that (I think) I better understand where its proponents are coming from. It appears as if the main motivation for bestowing such importance on the notion of representation is its role in defending a certain view of what it means to be human, namely a being that is free to act and contemplate. Nevertheless, by inventing the notion of representation I think we are actually doing ourselves and this view of human agency a disservice, especially because it covers over the fact that the traditional mechanistic view of the mind leaves no proper room for this conception agency.
However, without the representations to account for why we act for reasons, the mechanistic view is doomed to self-refutation. It would deny the basis for our accepting it as a valid and viable argument – our acceptance would just be contingent on some arbitrary mechanistic operations. Moreover, if we dispense with representations, this would not only invalidate the mechanistic view of the mind, it also casts doubt on the wider mechanistic worldview of which it is a part. This is why rejecting representations is not an option for someone committed to this perspective, no matter how untenable the notion is.
In conclusion: I believe that the notion of representation attempts to make space within a traditional mechanistic worldview for a genuine intuition about what defines us as human agents, but that this attempt is doomed to failure. To some this conclusion might sound extremely undesirable, but I suggest that this failure should rather be seen as an opportunity to gain a better understanding of what it means to be a human agent.
I submit that it is only by pushing the traditional mechanistic worldview to its own radical self-refuting conclusion (rather than attempting to make it more bearable through ad hoc theoretical postulates), that we can begin to focus on moving the scientific debate forward.