Cognition after the representation war (part 1) — An epistemological placeholder
New approaches in cognitive science are finally overcoming the restless debate about representation in human cognition. We retrace the evolution of the concept of mental representation until the current reappraisal of its centrality.
Traditional cognitive science, in which the classical cognitivist paradigm and the more recent connectionism have been predominant, provided an overview of cognition fundamentally based on three concepts: mental representation, computation and information-processing.
Representation is the foundational concept of cognitive science, and despite it turned to be very problematic and an endless source of disagreement about its putative existence, constitution and role, it has never been abandoned.
Cognition has been mainly described as a mechanism consisting in processing cognitive states/structures, which are supposed to be representative of (standing for) perceived objects and events of the world. Making extensive use of the computer analogy, cognition has been depicted as an information-processing machine, in which representations are computed according to “formal rules”. With the progress of neuroscience, the search for those “rules” has been focused on neurobiological activity, stressing the idea of a brain-centred and brain-bounded cognitive system.
The notion of mental representation gained great popularity even outside the academic world, becoming a term of common use to refer to the “contents” of mind.
In its common sense, ‘mental representation’ is used for referring to the correspondence between a certain dimension of the cognitive system (e.g. brain, mind, language, memory, thought, etc.) and a phenomenon, which most of the time is a distal physical object or event, but may also be a past, non-existent or abstract element, namely, a physically-not-available item. Therefore, representations are generally considered as a sort of individual, inner acquisitions/reconstructions that we acquire/produce in order to be processed and memorized. In their most abstract notion, they are characterised as images or semantic items (e.g. ideas, concepts), while in their neurocomputation-like conception, as patterns of neuronal connectivity corresponding to certain perceptual experiences.
In the academic context, most of the theory about cognition is built upon the assumption about the existence of some sort of mental surrogates of perceived objects and events, despite a solid consensus about their supposed attributes has never being achieved .
Academic research about representation has been mainly driven by three general interrogatives: is there is any such thing as mental representations, if so, what are they and what is their role in cognition. Depending on specific approaches, hypothesis and theories have been formulated mainly in two forms of explanation: in constitutive terms — aiming to identify representations as states or structures, trying to understand what they consist of, for instance associating them to neurobiological structures — or in functional terms — trying to delineate their procedural role, describing them as processes-functions performed by the brain. Researchers, encouraged by the progress of brain study techniques (e.g. fMRI, EEG, rTMS, Optogenetics), had produced theories mostly following an experimental approach, formulating models to be used as explanation of behavioural and psychological phenomena, describing representations as logical propositions, rules, concepts, images and analogies, to be tested through (or deduced from) observation of the brain activity during task-based experiments [3].
The concept of ‘representation’ has been introduced as “epistemological placeholder” and semantic tool, necessary for developing models of explanation about cognition. Over time, the concept went beyond its mere analogue explicatory function, becoming the core of the scientific-philosophical investigation about the relation between neurobiological activity and subjective cognitive experience — beyond the “here and now” — in view of understanding the emergence of faculties like consciousness and semantic reasoning.
The adoption of ‘mental representations’ stems from the necessity to explain all those situations in which the agent is able to think about something which is not currently present in the environment, so which cannot be perceived using sensorial stimuli directly collected from the world. In order to develop hypotheses and models about representation-hungry cognitive faculties [2] such as memory, imagination, mental object manipulation, planning (etc.), it was necessary to introduce a cognitive item standing for the absent object. In a broader sense, representations fill the ancient epistemological gap between physical-chemical environmental events and neurophysiological human activities, along the continuum by which complex cognitive functions emerge.
We can maliciously describe the current discussion about representations as a discordant collection of answers to a discordant collection of questions, without an actual consensus about the terms of the issue.
The lack of consensus about the form and status of representations has led to great terminological and semantic ambiguity. Indeed, different writings could use the term into account with different connotations, making works difficult to correlate and generating ambiguities in terms of levels of explanation. All this, in addition to the rising criticisms about the actuality of representations at all, generated what Clark ironically called representation war [1], which we can quickly illustrate as the conflict between representationalist and anti-representationalist approaches, where among the first we have the heterogeneous set of theories defending (or depending on) the “ontological factuality” or at least the epistemological necessity of mental representations, while among the latter we find representation-sceptic arguments or theories which are not compatible with any form of mental representation. For a deepening discussion about the theme of representations in relation to enactive cognition, Daniel Williams’s work provides an accurate analysis [4][5].
[1] Clark A., Predicting peace: The end of the representation wars. Open MIND. Frankfurt am Main: MIND Group; 2015 Jan 15.
[2] Kiverstein J.D., Rietveld E. Reconceiving representation-hungry cognition: an ecological-enactive proposal. Adaptive behavior. 2018 Aug;26(4):147–63.
[3] Thagard P., Cognitive Science, The Stanford Encyclopedia of Philosophy (Spring 2019 Edition), Edward N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/spr2019/entries/cognitive-science/>.
[4] Williams D., Predictive processing and the representation wars. Minds and Machines. 2018 Mar 1;28(1):141–72.
[5] Williams D., The Mind as a Predictive Modelling Engine: Generative Models, Structural Similarity, and Mental Representation (Doctoral dissertation, University of Cambridge).
