Andy Blunden, August 2010

Where we are now with Concepts

Thought-Forms and Mental Images

Robert Brandom has correctly drawn our attention to the two sides of a concept, that described by Representational Theories of Mind and that described by his own Inferentialism. It seems to me that he is correct in this. Both what constitutes the object’s existence and what constitutes the object’s significance must surely both be contained in anything we call a concept, but it is the meaning which is specifically human and conceptual. But let us reflect for a moment on the idea of concepts as representations.

A concept is often understood to mean a thought-form or ‘mental representation’, and whatever else it may prove to be, it must at least be this also, for otherwise we would all be talking at cross purposes. However, there are a number of problems with the very business of making thought-forms, in themselves, objects of discussion and investigation. Also, not everything that passes through our consciousness counts as a concept, and even amongst what does, not all can count equally as concepts. As Robert Brandom says, the concept of concept is a normative concept.

I understand by concepts thought-forms which meaningfully represent something in the world, so we need to work out what it could be in the world which a concept can ‘represent’. To be a ‘representation’, a concept must represent something, whether well or badly. So concepts are in some way a more or less adequate ‘reflection’ of something which could be in the world outside thought. I say this fully cognisant of the limitations of the idea of concepts as ‘reflections’ of the material world. But there is a reality criterion which is relevant to all forms of consciousness and applicable to all forms of life from amoebae to human beings: consciousness is part of an organism’s struggle for survival in its environment. Only insofar as its consciousness provides an adequate basis for the organism’s activity in the surrounding world can the organism survive and flourish. Otherwise neither the organism nor its ‘thought-forms’ are viable. It is solely in this sense that I use the term ‘reflection’ to describe thought-forms. A map of the world is a reflection of the world, but only to the extent that it can be used as an aid to navigation; in other words, to the extent that it means something.

The ‘world outside consciousness’ must be understood as some collective form of life, for it is only in relation to the life-activity of the organism that a concept has any bearing and can have any real meaning. If a certain concept reflects some material object, just what aspect of the object is reflected: its weight, the number of its atoms? Concepts do not ‘reflect’ the infinity of such attributes of an object, but only those that are in some way relevant and meaningful in relation to a specific interaction with human activity. Just as a microscope or a telescope, a piece of litmus paper or a radiation detector, each reflects objects in its own way, concepts reflect properties of the natural world, only insofar as those properties are manifested in human activity. So there is a real problem in taking the representation of an object given by a concept to be any kind of sensuous image, for there is in general no sensuous form identifiable with a concept, except that narrow class of concepts (the colours, etc.) where this is trivially so. Sensuous forms provide clues as to the identity of the object in the same way a detective collects clues to the identity of the criminal.

Consciousness mediates between an organism’s behaviour and its own physiology. The organism’s behaviour and physiology are part of the material world, but consciousness relates to the world only mediately, via the organism’s activity. The idea of direct, unmediated contact between a thought and the material world outside consciousness is incoherent and meaningless. A concept is meaningful only insofar as it is in some sense, direct or indirect, a guide to reality in terms of what matters, in terms of the organism’s real-life activity.

Even though we know that dreams serve an important role in maintaining our health, we don’t call the forms of consciousness we experience while dreaming ‘concepts’. Nonetheless, it doesn’t matter how weird or contradictory a concept may be, whether it belongs to religious fanaticism, superstition or a computer game, those thought-forms which are part of how people organise their own activity within some collective form of life, count as ‘concepts’. Doubtless it is impossible to draw a line between some chimera we experience during a drug-induced delirium and the concept of alien abduction, but there is nothing to be gained from exploring such elusive border lines. The point is that concepts are not just any mental phenomenon, but those which are an essential part of some collective form of life.

I am talking about human life. Some of the most elementary forms of conscious activity are available to animals, and even machines can effectively mimic some forms of human action. But this is not about rat-racing or computer programming; it is about human life. There is always room for surprises here, and whether or not a chimpanzee has been observed using conceptual thought is not a question I will be touching upon here. But surely conceptual thought proper is available only to human beings. The very primitive forms of syncretic thought characterising, for example, the normal mode of conscious activity of infants, qualify as ‘concepts’ only insofar as they constitute forms of conscious activity which are genetically connected to conceptual thinking or enter into conceptual thought properly so called as a component part.

So sensuous images are really a misleading way of thinking of concepts. A thought-form appears to be a very complex kind of thing, more to do with symbols than images.

Let us accept the idea of a material world given to consciousness through sensation. But this just moves the problem of forming concepts of things one step downstream. The problem of making sense of a stream of electrochemical impulses on the nervous system is essentially no different from the problem of making sense of the electromagnetic and mechanical vibrations impacting on the sense organs. If we accept the idea of the senses giving us a mental picture of things in the world which we can sort into categories, then in effect we are suggesting that there is a little human being, a homunculus, sitting inside the brain cavity, watching a kind of TV screen and taking notes. But how does the homunculus understand what it sees on the screen? Does it have an even littler homunculus in its own head? And it makes no difference if you call the little TV screen “neurons” and the homunculus a “control centre,” (Damasio 2003: 207-8) or “centralised strategic processing mechanism” (Barsalou 1992: 90), the problem is the same.

A camera represents but it does not understand, and surely if we are talking of concepts it is understanding that we are concerned with. A concept is not a material thing, and nor is it a pattern of neuron activity. The whole idea of a concept as a copy of something in the world is quite incoherent from the point of view of science. If concepts are mental representations of categories of things in the world, what aspects of reality are represented? who looks at and interprets the representations? And how?

How can a thought-form be identical to (or ‘map on to’) a material object? The very question poses an impossible dichotomy, but how else can the idea of truthful reflection of the world be made sense of? In fact, concepts cannot be understood as mirror-images of Doppelgänger in the material world, but only as entities which span both worlds. Concepts must be both mental and material. Instead of looking for matching pairs, we should look for entities which by their very nature transcend the mind/matter dichotomy by participating in both thought and matter. For this, we must turn to the real forms of mediation by means of which human thinking is connected to our material environment. So, two interrelated kinds of entity hold the key: artefacts (including words) and activity (including speech).

Artefacts are all those products of human labour which are objectifications of human needs and aims, materialisations of thought – words (whether spoken or written), tools, machines, buildings, books, movies and even our own bodies: objective because they are material objects, subjective because they are what they are only thanks to their place in the social life of human beings. Activity is all the systems of purposive social action by means of which our thoughts are manifested and our needs satisfied: subjective because we mean them, but objective because they happen in the material world. Actions are always mediated by the use of artefacts, which are themselves products of social activity, given meaning by those same activities.

But this still leaves unresolved the problem of how artefacts and activity are correlated in a concept, of exactly what can exist both subjectively and objectively to be constituted in consciousness as a concept. Situation meets these requirements, or we could say ‘problem-situation’, or ‘predicament’, but I take it as implicit in the idea of ‘situation’ that it represents problem at some stage of gestation, crisis or resolution, which constitutes the subjective significance of some conjunction of relations. Drama also provides us with notions of scripts, scenes and other event, story or character schemas which are also subjective-objective entities. It is important to highlight the special ontological status of these subjective-objective entities, because insofar as concepts represent, it must be these which are represented.

Consciousness is part of the existence of a situation too. Concepts, or situations, are both subjective and objective. When we talk of concepts as representations, images or whatever, as mental objects of some kind, this is a kind of reification. We cannot in principle observe our own consciousness, let alone forms and shapes within it. Consciousness is the observing, a process – not the observer or the observed, but the observing. When I look at something it is that thing that I see, not a representation of it. An illustration of how implausible the idea of an ‘internal image’ is, is given by Barsalou:

... categorical knowledge is grounded in the brain’s modal systems rather than being represented amodally in a modular semantic. For example, knowledge about dogs is represented in visual representations of how dogs look, in auditory representations of how dogs sound, and in motor representations of how to interact with dogs. Because the representational systems that underlie perception, action, and affect are also used to represent categorical knowledge, the conceptual system is neither modular nor amodal. Instead, perception and conception share overlapping systems. (2008: 92)

Insofar as a human being forms a representation of something, it seems inescapable that the whole of our body, and the artefacts we use, are entailed in forming any representation.

Human beings are born realists. We act as if the things we perceive exist independently of our consciousness and activity. Only later, thanks to critical reflection, do we come to realise that the concepts we form of things in the objective world are actively constructed from material belonging to our own culture and activity, whilst other people may think of things differently, through the lens of their culture and activity. In the same way, we are also born realists with respect to our self-consciousness. Even if we accept that our concepts are not copies of things in the material world, we tend to naively take concepts to be entities ‘existing’ in our mind. But as we have seen, this cannot be the case. We may learn to take a critical attitude to both the content and form of our own consciousness. Ruth Millikan (in Margolis & Laurence 1999: 537) tries to avoid reification by saying that a concept is a human ability or capacity rather than an entity. But as Anna Sfard (2008) convincingly shows, abilities and capacities are themselves products of reification, giving seeming permanence to what may be subject to unnoticed situational variation. Linguists avoid the problem altogether by treating concepts solely in the form in which they are objectified in language, whilst sociologists and behaviourists avoid the problem by concerning themselves only with behaviour and treating consciousness as a ‘black box’. But these routes are not available to psychology, which must perforce make consciousness the object of its investigation.

Let us very briefly review what is known of visual images. Physiology tells us that nothing like the image projected on a movie screen is provided to the brain by our eyes. The work of V. P. Zinchenko (Zinchenko & Vergiles 1972) has established that visual perception is constructed from involuntary saccadic eye movements of 20-200 msec in duration and it is only change in the stimulation of the retina which produces a visual sensation. Our visual awareness is a highly constructed experience. The first step an infant takes towards human cognition is when it is able to intervene in the intense stream of incoherent visual (and other) impressions and momentarily prolong an impression which stands out for some reason. This infant eidetism, the prolonging, or freezing, of a sense impression beyond the time when the stimulus is present, is the first manifestation of control over perception, and the first departure of sense experience from what is immediately given from the environment. But what we talk of as an image is here just an after-effect of the action of the environment on the sense organs. Later this relative independence from the immediate stimuli allows us to isolate separate aspects of the stimulus and combine them in different ways, more and more under the control of the individual. None of this is conceptual thought, but this active attention to stimuli does lay the foundation for conceptual thought. But thought is conceptual to the extent that it is freed from the immediate situation.

We know from neuroscience that the shape, speed, trajectory, colour and meaning of a moving object are perceived by distinct neuronal modules. There is no point in the brain where the various streams of sense data are brought together to form an integral image, any more than an image exists at a point behind the mirror in any material sense. And whatever is involved in thinking of a concept, it always entails a word or some other artefact which symbolically represents what is universal, whether or not an exemplar of the object is also involved. Concepts may entail mental images as well, as part of a more complex process. Concepts can only be made sense of by taking nervous activity to be just one aspect of a concept, along with the artefacts implicated in a person’s actions and the activity of which their actions are a part: “coupled systems,” to use Nancy Nersessian’s term.

Networks, Plots, Categories, Theories and Institutions

The taking of concept-formation as a process of categorisation is not necessarily linked to the taking of concepts to be images or representations of the object. Nor is categorisation necessarily linked to the presumption that the criteria for categorisation are contingent attributes abstracted from the object. Both representationalism and anti-essentialism are problematic aspects of the cognitivist conception of categorisation. But an important feature of categorisation is simply that it is a generalisation. Categorisation is one possible means of linking an object to other objects and specifically to objects of broader scope and significance, to a conceptual context. At the end of a process of categorisation, in which a concept is successively stripped of contingent attributes and concreteness, one would arrive only at the completely empty concept of ‘everything’. But the first step of categorisation is always meaningful, even if categorisation cannot by itself exhaust a concept.

You may not have a concept of an ‘armoire’ but knowing that it is an item of furniture would surely be a step towards and indeed part of the concept of armoire. A number of approaches to concepts suggest different steps which have the function of connecting the object into a wider concept in some way. For example, I have mentioned the theory theory which rests on the fact that concepts are always units of some theory, a general idea about some domain of phenomena, whether formal or informal.

Then there is the idea of a semantic network in which the positing of any one concept activates a range of related concepts. For example, ‘spoon’ activates a semantic network of kitchen things, so that ‘knife’ is immediately taken as a kitchen knife. The idea of semantic network is a generalisation of a range of types of association, among which one could include the theory theory as well as the network created by any institution (cuisine) or complex object (kitchen). For example, you may not know what a carburettor really is, but you may know it as an automobile component, and that is part of the concept of it.

If we take an ‘activity’, to mean an aggregate of individual actions which has an on-going existence in society at large, in pursuit of some objective, then subsumption under such an activity, of whatever kind, serves to connect a concept into some definite series of other, related concepts. Carburettor and fan belt are not just associated because they are both car parts, but each plays a very specific role in effecting the correct functioning of an internal combustion engine. The same goes for the theory theory. Likewise, accountant and cashier are not just associated by both being bureaucratic roles, but have specific relations within the division of labour of bureaucratic organisations.

Taking concepts as indicating some definite connection with a social practice has the same function of explaining how one concept sets the scene, so to speak, for another, but offers the prospect of contributing more to a concept than its simple association with other, related concepts. The narrative approach goes one step further in as much as it organises concepts into a specific structure, namely a plot arising from a predicament. Metaphors and analogies also link a concept to other concepts, though in this case the linkage is imaginative rather than contextual.

The question remains as to whether these forms of association, connection or contextualisation have any psychological reality. Does the narrative from which we learn about ‘refugee’ have any psychological reality? Does our practical knowledge of the workings of internal combustion engines have any relevance to the concept we have of ‘carburettor’? Does our knowledge of cricket have any relevance to what we make of the concept of ‘keeper? If so, is there more to association than frequencies on a table of word associations? Does the practical intelligence acquired in some field of practice have a real effect on how we conceptualise the situations and objects we deal with in that practice?

Jean Mandler proposed scripts or story schemas (1984) or schemata (Neisser 1967) as a basic mental structure, corresponding to patterns of action, as units of human life and the foundation of concepts. These concepts have been used in a great deal of psychological research and therapy, and provide a rich source of understanding. My concern is that these ideas represent only the ‘molecular’ level of situations and relationships, whereas concepts represent the molar or broader motivating context of the situation.

I think this is the main problem, if and how the concept of something connects up, psychologically speaking, with the broader meaningful setting in which the situation arises, and if and how a concept draws real content from such connections.


Cognitive psychologists have determined that representations of individual, idealised exemplars of objects instantiating a concept play a role in the psychological reality of concepts. Further, it has been found that the ideological formations which determine people’s expectations in relation to their experience also play a role in how people grasp the world around them. People call on beliefs about the defining features of kinds of object when categorising them or recognising them. But how individual representations interact with ideological contexts is not so clear. Cognitive psychologists have also disproved a number of preconceptions which may have been taken for granted within analytical science, and in so doing, have generated a very helpful dialogue about the psychology of concepts.

The test procedures used are invariably based on the untenable assumption that some invariant mental representation, called a concept, exists, which may be elicited by any test procedure without the test procedure itself being a determining factor. But if we are to make any sense at all of the idea of a person ‘having’ a concept of something, then that concept has to be admitted to manifest itself in a wide variety of different behaviour under different test conditions. But how to represent a concept in such a way that responses elicited in a wide variety of test conditions may be unified as realisations of one concept?

Robert Brandom has demonstrated that representational theories of the concept tell us only one half of the story, and the less essential half at that. Concepts testify to the connection of some specific situation to a set of inferences, and it is these inferences which the concept associates with the conditions of its existence which is characteristic of conceptual thinking. A concept registers a predicament and its resolution, coming to us as the ‘lesson’ of a story.

Linguists have determined that the cultural world, with its mass of artefacts, especially language, and our collaborative use of these cultural products, is crucial to if not exhaustive of the formation and use of concepts. Some linguists also believe that our comprehension of abstract concepts is built upon more visceral and immediate practical experiences, by means of metaphor and mental models of situations.

It is almost impossible to make generalisations about the diversity of approaches to the study of concepts which I have reviewed. But this in itself speaks of the main challenges facing the study of concepts. The dominance of analytical science and philosophy over the past century has brought a number of problems to the study of concepts. The central characteristic of the analytical method (which is also its great strength) is the segregation of disciplines. This goes to the extent that those who follow the analytical method in psychology, such as the Cognitive Psychologists considered above) may be ignorant of work of analytical philosophers and sociologists, and vice versa. On the other hand, philosophers in the ‘philosophy of mind’ do not offer a critique of analytical science at all, but rather constitute a sub-discipline within analytical science. So, the separation of experimental science from genuine philosophical critique has not only allowed naïveté to prevail in the Psychology of Concepts, but has promoted the fragmentation of the study of concepts between numerous disciplines.

Difficult as this situation has been in all domains, it has created an impossible situation for the science of concepts, because concepts are by their very nature subjective/objective entities. This not only means that concepts span both psychology and the social sciences, and require an interdisciplinary approach, but at the most fundamental level they violate the ontological presumptions of analytical philosophy. The very idea of entities which are both subjective and objective also violates the logical principles of analytical philosophy, which is wedded to formal logic. Formal logic obliges a dualistic approach to concepts which creates an impassable barrier to the solution of all the major problems confronting the science of concepts. We need to revive the art of thinking with true concepts rather than tick-boxes.

The conviction that concepts are the products, or outcomes, of a process of development which can be understood separately from the process which produced them has led to a situation in which many cognitive psychologists, linguists and learning theorists nowadays reject the very idea of ‘concept’ having a place in science. We urgently need an approach which grasps concepts as processes, not things.

To address these problems I shall make a detour into the history of philosophy where these problems were resolved speculatively, without recourse to the methods of experimental science.