A blog about problems in the field of psychology and attempts to fix them.

Sunday, August 21, 2011

What are Concepts? (Part 1)

Amongst my many pet peeves about Intro Psych textbooks is how they handle 'concepts.' It is not so much that they use the term particularly badly, it is that they do not provide any context (i.e., this is a problem even if we put aside the need to redesign Intro Psych more broadly). In particular, I don't see how 'concepts' make any sense without at least some discussion of 'percepts' - the two ideas are intimately intertwined.

Percepts are the things given by perception / taken in the act of perceiving.

Concepts are the things added to perception / things with the taken.

Traditional and Mainstream Contemporary Views
Most of modern psychology grows out of the tradition of "British Empiricism", at least in the sense that most psychologists assumed that perception gives some sort of atomistic units that must be put together and made sense of through some mental process. They assume that percepts - what is gotten from the world - are, as Gibson would somewhat mockingly say, a "scintillation of sensations", an inherently meaningless array of color points, pressure measures, scent elements, etc. Sometimes psychologists more generously start from something like the stationary retinal image, in which case percepts are a disconnected patchwork of colored shapes, sounds, touches, smells, etc.

From either of these starting points, the mind must provide relationships between the elements of sensation, and must provide meaning to clusters of those elements. This leads to the assertion, still occasionally found in the extreme form, that all relations are 'mental' and all meaning is 'mental'. By this understanding, concepts - what is added to the percepts - include any relationships we might express by a preposition (next to, part of, connected with, after, over, because of, etc.), as well as anything we might express in terms of category membership (chair, tall, smooth, dangerous, edible, disgusting, etc.).

I don't have a problem that Intro textbooks encourage this way of thinking, but I have a problem when they don't provide what is needed for students to think clearly along these lines. A few things that any exposition of concepts and percepts should make clear:
  1. Percepts are those things given directly, our most basic elements of experience. They are (psychologically) unmediated. 
  2. How we arrive at our concepts is very mysterious, and if you don't like having a separate soul-stuff or mind-stuff, then you need a theory about how the physical brain can 'add' to what it gets from the world. (Incidentally, this is what the computer model of the mind is about, i.e. modern cognitive psychology, but the part of the textbook explicitly about 'cognitive psychology' is never meaningfully connected with the part that talks about 'concepts.' Uhg!)
  3. What we think the minds needs to supply is therefore a function of what we think we can get from the environment itself. 
Getting to point 3 provides the minimal background needed to meaningfully discuss alternative approaches in later classes. What types of things might those later classes discuss? The works of William James and James J. Gibson provide good examples.

A bit of William James
William James's philosophy flips much of traditional psychology upside down, and asserts (usually convincingly) that the result is right-side up. Thus, it is a good place to start talking about alternatives to the traditional view. Many of James's criticisms involve the psychologists fallacy, wherein someone mistakes the endpoint of an process for the starting point of the process. I don't want to focus on the fallacy itself, but do want to explore one of James's critiques that involves the fallacy.

James claimed that the British Empiricists had it exactly backwards: The most basic unit of experience is the complex world in which we find ourselves - for example, right now my experience is of a room with a TV and two toddlers, a table, some chairs, a computer on which I am writing, etc. James points out that it is a complex and difficult-to-develop skill to "see" the world as composed of more basic elements - for example to see the table as composed of a brown circle connected to perpendicular thin rectangles. It is an even more sophisticated ability to "see" the world as 'in perspective' - for example, to see the table top as 'appearing as a brown oval' (because I am not viewing it from directly above).

This is not mere speculation on James's part, and one does not need to buy into his 'world of pure experience' to be convinced he is correct. Any study of the history of art, or of the development of artistic abilities within an individual, would instantly evidence that the ability to see the world as composed of basic elements, and especially to see it as 'in perspective', is a sophisticated skill. There would not be a thriving market for art books that show us the world broken up into basic shapes if everyone already saw the world that way with ease.

Although he did not bring up James, or the psychologists fallacy, Tom Stoffregen gave a heartfelt, incredibly serious, and also hilarious, talk on this subject at this years ICPA meeting entitled "How Plato is used to Brainwash Helpless Toddlers." In it Stoffregen showed examples of so-called "educational programs" designed to teach toddlers to see the world as composed of basic platonic elements. Imagine traumatically Disney-esque songs about how a person is just a circle on top of some rectangles. Tom felt that this was incredibly problematic, because he thought it was part of a vast conspiracy to make it difficult for his undergraduate students to understand what how perception really works (i.e., to understand perception from an Ecological perspective). It is... or at least it is part of a vast network of early childhood experiences that have that effect.

Of course, James would not label it problematic that people develop the ability to experience objects as composed of 'platonic elements', but James would have agreed that it is very problematic that people do not realize how hard it was for them to master this ability. By not examining the development of this skill, psychologists fell into the psychologists fallacy, they assumed that a sophisticated skill was what we started with, they assumed erroneously that the most basic state of vision is to register our world in terms of colored platonic forms. It is true that we can take the things we experience and break them apart into simpler pieces, but we are getting everything backwards if we thereby assume that we started with the smaller pieces!

In none of this does James reject the original meaning of the terms 'precept' and 'concept', but he suggesting that we have the contents of the categories backwards. James is suggesting that what we get directly is a world full of meaningful objects and events in relationships with each other (i.e., a 'stream' of things and happenings). He is further suggesting that from those direct experiences we 'mentally' create the categories of 'color', 'shape', 'element', etc. Thus, I perceive the table, and I sometimes conceive of it as a collection of connected basic shapes.

(For now I will skip over how the more eccentric stuff James wrote later in life alters this story a bit more.)

Ecological Psychology
A few people have noted that Gibson's work can be best understood within the intellectual lineage of 'American Philosophy'. That is, Gibson's work connects strongly with James's work (see discussions by Harry Heft, Ed Reed, and a few other authors in my almost-out edited volume). One way to explain the relationship between is to say that Gibson was uncovering a mechanism by which James's claims could be true. That is, Gibson's theory takes as a description that people directly experience the objects and events of the world as interconnected and meaningful. Gibson claims that the organism 'resonates' with complex patterns of energy in the environment that are 'specific to' the objects and events of the surrounding world. Because objects and events have consequences for an active organism (in exactly the way that a disjointed scintillation of lights and sounds do not), what the organism 'gets' is inherently meaningful, in at least a minimal sense.

If it is true that we are sensitive to such complex patterns (higher-order invariants), and there is plenty of evidence that we are, then we can 'get' many (if not all) relationships between objects and events in our surrounding directly through perception, and 'get' many (in not all) meanings of those objects directly through perception. Thus 'percepts' can include relationships and meanings.

It is a problem when 'concepts' is taught without being contextualized by 'percepts'. It is problematic because the term makes no sense except as part of a pair, and because a failure to understand the relationship between the pair of terms makes it very difficult to discuss alternative ideas about what those categories should include. This has the most obvious effect of leaving students with an incomplete and disjointed understanding of the mainstream system they are being taught - some variant of modern cognitive psychology. It also causes the less obvious (to most people) problem, that students have tremendous difficulty understanding alternative approaches to mainstream psychology, including some of those rapidly gaining traction (i.e., variants of 'embodied cognition' and 'extended cognition').


Brief Connection with Future Discussion
Because the traditional distinction between precepts and concepts has been around for so long, people often think about challenges we face (both in our normal lives and in the lab) as being either perceptual or conceptual challenges. For example, a given experiment may be classified as either a 'perceptual task' or a 'conceptual task', based on whether we think people can succeed at the task based on percepts alone, or if we think it requires the use of concepts. This leads to much confusion when alternative approaches are discussed. For example, an essential goal of many ecological psychologists is to show that so-called 'conceptual tasks' can be solved with perceptual mechanisms. This quickly leads to a discussion in which the labels create unnecessary confusion. Everything essential in the declaration that "conceptual task X has been solved by perception means", is more cleanly expressed in the assertion "X is a perceptual task." But the latter claim is deemed a priori invalid by the mainstream. The battle for clarity thus often gets hijacked by the historic momentum of our terms.


  1. "Any study of the history of art, or of the development of artistic abilities within an individual, would instantly evidence that the ability to see the world as composed of basic elements, and especially to see it as 'in perspective', is a sophisticated skill. There would not be a thriving market for art books that show us the world broken up into basic shapes if everyone already saw the world that way with ease."

    Well said! As someone who studies categorisation, I have this argument quite frequently. Most people think that we first perceive an object as an individual entity and then use some similarity/classification process based on prototypes/exemplars/global similarity/essences, whatever, to place that entity into a category. Rather, it seems to me that most categories are in face perceptually constrained in a way that allows direct access to meaning. The problem is that the "features" around which these categories cohere are complex, multidimensional, and not easily expressed in language. The features that we can easily express in language (e.g., colour, basic shape, number of legs, etc) tend to be indicative, but ultimately non-diagnostic of category membership.

  2. Titchener apparently used to show people a chair and ask them what they saw, then yell at them until they told him about the primitive features he thought were the basis of perception :)

  3. Sabrina,
    Psychologists' willingness to dismiss the environmental support of psychological processes is always fascinating. It is often very difficult to investigate, or even talk about as you point out, and so I understand why people don't want to do the work. But it is another problem entirely when they don't even see that there is work to be done.

    P.S. Oh that rascally Titchener!

    To be fair though, he had one of the few theories of introspection (that I know of) which asserted significant training was needed to introspect correctly. Maybe he was just trying to give his subjects a free taste of graduate education :- )

  4. The most basic unit of experience is the complex world in which we find ourselves

    As an example of James-style composite percepts, wouldn't it be more instructive to start with the world as seen by babies rather than as seen by trained psychologists? In "Scientists in the Crib" (which I'm now reading) they claim that babies initially show the most interest in stripes, or more generally, edges. Thus, I would guess that what babies "perceive" is, at most, a surface something like a Hans Hoffmann painting (see


    for an example) and possibly even something more primitive like a paint-by-numbers version with the "colors" of phenomenal experience replaced (in essence) by numbers representing distinguishable patterns of neural activity. Ie, they initially might experience merely distinguishable areas of the FOV but at first not experience any area as comprising a separable entity. Then, separable entities - ie, "objects" - might emerge as a baby begins to notice connected edges and/or "colored" areas that move as a unit relative to the mostly static "background". Pretty ecological, no?

    Question: What Eric calls the "traditional" concept-percept scheme and the scheme Sabrina says "most people" assume (category-percept) are reminiscent of Kant's conceptual scheme-perceptual content (AKA, intuitions) distinction except that my understanding is that the latter requires language. Which is why I balk at attributing the "concept" of food to the rat (but speculated that we were perhaps assuming different definitions for "concept").

    Loosely related question: would people formally educated in cog-psych in general (ie, not just those specializing in language) typically have been exposed to the writings of Donald Davidson - who challenged the scheme-content distinction?

  5. Charles,
    Easy parts first...
    "Loosely related question": No, most cognitive psychologists would not have been exposed to Donald Davidson. Of course, I would feel comfortable claiming that here are no philosophers to whom most formally educated cog-psych people would have been exposed (Jerry Fordor and Daniel Dennett are widely read, but still 'most' would be generous). Psychology has a very short memory span, and a very narrow gaze.

    "Question": You are right to see a connection between the modern cognitive approach and Kant's work. Modern cognitive psych (intellectually) grows fairly directly out of the 'continental tradition' in philosophy, and of course Kant is a key player there. Again though, that should not be taken to imply that most cog-psych people know anything about Kant, Kant's views, or why Kant held those views.

    The relationship between language and concepts would be a much longer discussion. In particular, we would need to be terribly careful about what we are trying to explain when we talk of concepts and how we are trying to explain it. Unless concepts are made of language-stuff, I'm not sure how we could a priori rule out concepts without language.

    I'm not totally against your description of infant's experience, but, either way, the question is where we go from there. The cognitive psychologists want to talk about infants mentally putting things together. The ecological psychologists (and behaviorists) want to talk about an expansion of the variety of things infants respond to, and a specialization in how they respond to different things. Does that make sense?


    P.S. You were write about the misspell in the previous post, the main posts can be fixed, even though the comments cannot.

  6. Eric -

    Tnx for the replies. I'm a little slow responding because of HDMI problems (hurricane damage mitigation initiatives). (:>)

    re DD: that's what I would have guessed. I had the good fortune (IMO, anyway) early in my foray into philosophy of stumbling upon Rorty, and thru him Sellars and DD (et al, incl Wittgenstein) and have found them very helpful in trying to understand some of the issues you, Andrew, Sabina, and Ken Aizawa address. Especially important, in my view, is the social aspect of issues that are often seen as relating only to individuals, a common theme among Rorty, Sellars, and Davidson. I appreciate that it's not possible to study everything/everyone when executing a formal program, but I have been surprised how few among the people I encounter on blogs who exhibit intense interest in "the mind" have even heard of them. They are, of course, rather dated but I regularly encounter current discussions of supposedly "open questions" that appear to me (possibly mistakenly) to have been more-or-less resolved decades ago by one or more of those folks.

    As for the infants, so far I have almost no grasp on what "makes sense". The child dev psych books I'm reading attribute mental abilities to babies that seem quite surprising, attributions based on experiments their cursory descriptions of which sound reasonable. But another thread that runs through Sellars and Davidson is that the best answer to various questions about temporal priority in the development of mental abilities is likely to be "neither, they develop simultaneously". But I'm confident I'll get helpful insights on those issues in these venues.

    re concepts and language: I eagerly await posts from you and Sabrina on that issue, my primary interest.