Associations are principles whereby impressions come to exist in different capacities than by what was received initially. So for example, in having the impression of brown, furry, smelly, and so on by association we come to correlate these actually distinct impressions into one identity known as a dog. Another example is the formation of concepts, which are universal in a way that singular impressions could not be. Now, granted that impressions are in themselves truly distinct, it can be established that the process of knowledge is not merely passive, since some active agent is necessary in order that distinct impressions can come to be comprising thought that is unitive and universal. But this means only that this active agent is directly acting upon the received impressions in what is a typical instance of change, which is in the traditional sense the actualization of a potential. That is to say, our received impressions have the potential to be understood and formed in universal, transcendent terms (Hume's notion of causality won't do here since the change involved here is not a matter of relations between impressions, but is rather about a necessarily connected operation occurring between impressions themselves and some active agent). Now as such, any actualization of a potency can only occur if that which is in potency is actualized by that which is actual. Thus, we must submit the existence of the self, which we call that which is 'actually associative.

However, is this correct? Is there a way of approaching the reality of an active agent without necessarily arriving at the need of the existence of that which is simply active, that which is responsible for the occurrence of the change taking place in our processes of association?

3 Answers 3


You are describing the weakness of Hume's account that Kant pointed out. Hume held that impressions associate, or appear to do so, just in virtue of their occurence, i.e. the stream of impressions acts by constant lumping together of properties. Under this passive receptivity picture Hume does not indeed need a self to play any active role. However, activities of the mind are far too complex to be accounted for by Hume's associationism, if universal terms are to be had at all they must come from elsewhere. As Pippin puts it in Kant on Empirical Concepts:"Contrary to Hume, it is not the case that impressions just by their occurrence generate a feeling of associability. Perhaps Kant’s most decisive objection to Hume is his claim that it is the mind which must actively order and associate them, and this according to acquired rules".

Whether we agree with Kant that the rules and universals come from the mind itself, or hold like Aristotle that they are somehow "contained" in the transcendent the hard question remains, how are they acquired? It is one thing to dismiss Hume's associationism as too naive, it is quite another to give a plausible account of how exactly mind extracts these regularities from the manifold of impressions. Saying that an "active agent" or "self" does it is as vacuous as vacuous can be. That simply creates the homunculus, the little man in the mind, and moves the problem on to the homunculus. Both Hume and Kant realized that, and tried to give a non-homunculus account, but Hume's does not measure up, and as Pippin summarizes: "Kant did indeed attempt to break quite radically from the Cartesian picture of knowledge as a kind of clarity or ‘glow’ of representations, or as an immediate grasp of some content, in favor of a thoroughly discursive picture of conceptual activity... However, I am also arguing that, just to the extent that Kant was successful... he raises what will become the problem for such an account, the ‘ground’ for this activity, a problem which he does not seem to be able to solve within the limits of his methodology".

We do not have an independent account of what "self" is and how it does what it does even today. That some "self" is necessary for activities of the mind is true enough, but that simply rephrases that mind is active, i.e. places a label on the unknown.

  • Perhaps my use of the word 'self' contained too many connotations. Your points are sound, and I think they agree with what I was most specifically suggesting in my question; namely that some actually active, capable 'mind' is necesarry if associations are possible, granted that associations are merely associatings, and associating is an action requiring an actor (that which is the actualizer of association, or the source of power in thought).
    – Chosen One
    Commented Feb 11, 2016 at 21:52

As Conifold points out, this is one of the problems in Hume's "associationism" addressed by Kant under the grand terminology of the "transcendental unity of apperception." Which he analyzes into sawdust.

Though I have not dipped into Hume in a long time, it may be that he felt no need to specify further what "sort of thing" this principle of "association" is on the formal model of Newton's "gravity." At the purely nominal level we could call anything the "self," of course. But the term carries with it not only the idea of agency... but also of "simple" substance in the Leibnizian sense. I think it was this idea Hume was attacking, and the very idea of "simple" unitary, irreducible groundings.

In general at this time, the idea of searching out some unitary substance was being replaced with "relations" between parts, with discursive, relational, or dialectical approaches. Rather than unitary "active agent" we have an "interaction" that unifies or mediates. In the Newtonian universe, for example, the "center" is now potentially everywhere and the "unity" is henceforth mathematical, with God for backup perhaps. Both Hume and Kant were struggling in some sense to generalize Newton.

Kant seemed to think he could "get to the bottom of it" by breaking up Hume's broad associational force into innumerable "subatomic" parts all interacting systematically. As a critique it is compelling. But there is no "one" indispensable agent, center, or simple substance. There is instead a necessary coherence and a kind of "boundary" in the noumenal. Actually, Kant does preserve and insist upon the simple, continuous "soul," but I think it was important to him that neither God nor this soul assume an explanatory function within his epistemology. Not sure, I could be wrong about this.

I'm not sure how one defines "active" at this level. Neither Hume nor Kant, already inspired by the physical sciences, will point to some "first cause." Today we have many other ways of dispensing with the unilateral idea of an "active" agent "responsible" for unifying percepts. We may speak of "responses" and "systems," "interactions" and "limits," or "regresses" or "attractors." On the information model we might say that the "most irreducible" thing we can talk about is an interactive, mediated (0,1) and never an active (1).

Even so, it seems to me there is always going to be some sort of necessary coherence or "self by any other name," (personally, I just think of it metaphorically as the enclosing exteriority of a "membrane") so that cognitive theories expunging the simple, unitary "self" or "soul" are usually renaming a more complex unity, which is in keeping with the move away from a unitary, theocentric model begun by both Hume and Kant.


Your questions operates with some terms from the philosophical tradition. They can be questioned in the light of neurobiological insight into the operation of the brain.

Such terms are “actualization of a potency”, probably taken from Aristotle’s metaphysics. As well as “active agent acting of received impression”, alike to a homunculus as Conifold states. And also the concept of a “Self”, a high-level construct which captures our conscious experience.

Ome of the core insights of neuroscience is the lack of an active agent. Instead, the prefrontal cortex collects a set of areas, which represent personal goals and other contextual information. They modulate competing representations in sensorial, memory related and motoric systems. (see Goschke, Thomas: Der bedingte Wille. In Roth, Gerhard; Grün, Kaus-Jürgen: Das Gehirn und seine Freiheit. 2009. German). But there is no single control unit. Instead there is a net of several competing systems, interacting by several loops of forward and backward information processing.

According to neurobiology, an area the “Self” does not exist and is not necessary.

  • According to Dennett, “Self” does not exist and is not necessary, but he is not a neuroscientist. The latter have a variety of opinions in addition to Dennett's, including accounts with coordinating "conductor" processes and/or areas in the brain, see e.g. Crick's Astonishing Hypothesis, which is skeptical of Dennett's "virtual captains". Computers do need central processor and operating system, which may be identified with aspects of "active agent" or "self", it doesn't have to be a ghost in the machine.
    – Conifold
    Commented Feb 11, 2016 at 1:26
  • @Conifold A neuronal net does not operate according to the paradigm of a von-Neumann-architecture. - Did the discussion between Dennett and Crick take place in the context of neurobiology or in the domain of philosophy of mind?
    – Jo Wehler
    Commented Feb 11, 2016 at 1:33
  • Astonishing Hypothesis is a book popularizing neuroscience, the framework is collective behavior of neurons in the brain (neural network would be a toy model). I quoted a passage on "mind's conductor" localized in the thalamus here philosophy.stackexchange.com/questions/31783/… Elsewhere in the book Crick comments on Consciousness Explained. But I think the "active agent" in this question does not have to be "consciousness" or "I", even lower animals may have active coordinated sensory processing.
    – Conifold
    Commented Feb 11, 2016 at 2:15

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .