A central thesis of Cognitive Ontology is that cognitive kinds are unlikely to reduce to neural kinds. I found one of the most exciting threads in the book to be an argument supporting this anti-reductionistic thesis, which Khalidi summarizes in today’s post, and which I’ll call “the Individuation Argument.” According to Khalidi, cognitive kinds are often individuated (at least in part) “externalistically,” whereas neural kinds are usually individuated “internalistically” (22-3). He suggests that this mismatch blocks the reduction of cognitive kinds to neural kinds.
The recurrence of Individuation Argument in a number of different chapters is one of the things I most enjoyed in Khalidi’s book. Still, I didn’t always find it convincing. Let’s grant for the sake of argument that neural kinds are typically individuated internalistically. I want to focus on the claim that cognitive kinds are externalistically individuated. I think Khalidi’s case studies show that the notion of individuation is slippery. And the sense in which some of the cognitive kinds he discusses are externalistically individuated is not one that prevents their identification with neural kinds.
Consider how the Individuation Argument is applied to episodic memory (Chapter 5). On Khalidi’s view, humans have a dedicated system for episodic memory whose adaptive function is to preserve information about past events for the purpose of registering exceptions to generic information stored in semantic memory. Khalidi argues that “if the episodic memory system is an evolved capacity, then it is individuated in part by the adaptive role it plays” – i.e., it is individuated etiologically (155). Episodic memory states are representational states produced by the episodic memory system. Khalidi claims they are “individuated primarily with reference to their etiology or causal history: They can be traced back to past episodes” (155).
I am skeptical that Khalidi has shown that either the episodic memory system or episodic memory states are externalistically individuated. Let’s start with the system. A cognitive system can be individuated by factors entirely “in the head” and yet also have an adaptive function. (Surely some neural systems, despite being internalistically individuated, have adaptive functions.) Even if Khalidi is right about the adaptive function of the episodic memory system, that does not mean it is externalistically individuated. What of episodic memory states? Khalidi characterizes episodic memory states as having a particular etiological profile: they have a “link to past episodes” (149). But he also recognizes that not all memories are factive (157). False memories are episodic memory states too, so long as they are produced by the episodic memory system. This shows that having a link to past experience is in fact not part of the individuation criteria for an episodic memory state.
Part of the problem is that Khalidi seems to conflate how a state is characterized with how it is individuated. Even if episodic memory states typically have a particular etiological profile, that does not mean they are individuated by reference to etiology. Khalidi also explicitly equates the identification and individuation of a kind. But these too are separate: the evidence that indicates a particular cognitive kind is present is not necessarily part of its individuation criteria. (My rumbling stomach indicates that I’m hungry, but I’m not hungry in virtue of my stomach rumbling. Stomach noises bear on the identification but not individuation of hunger.)
I suspect that there is equivocation between identification and individuation in the book’s discussion of concepts. Khalidi argues that developmental psychologists attribute richer concepts to children than is warranted by their “narrow responses and behavior” (60). He thinks psychologists are nevertheless justified in doing so for two reasons, one of which is that children are on a developmental trajectory that usually leads to their possession of adult versions of the concepts. Khalidi writes, “Rather than ascribe different concepts at every developmental stage, and attribute an entirely different conceptual repertoire at each stage, psychologists regularly say that the children possess these concepts, yet they have an incomplete understanding of them” (60). He concludes that concepts are individuated partly by causal powers and partly by etiology.
I didn’t entirely understand this line of thinking, but here’s one way of making sense of what’s going on. Imagine a child is responding to animals in distinctive ways that are suggestive of possessing the concept animal. According to Khalidi, her “narrow responses and behavior” don’t fully justify attributing animal to her. But the near certainty that the child will soon have a rich animal concept provides evidence that she possesses a version of the concept now. If this is what Khalidi has in mind, etiological considerations seem evidentially relevant to concept ascription, but they do not individuate the concept animal.
A cognitive state is externalistically individuated when factors “outside” of the thinker determine whether they are in that state or not. Holding fixed everything in the head, toggling facts about the environment or the thinker’s causal history changes whether the state is present or absent. I think Khalidi has shown that some of his cognitive kinds – such as myside bias – are externalistically individuated in this sense, but not all. His Individuation Argument is intriguing but its applications meet with mixed success.
Blogunuz çok güzel.. Sizi kendi sayfama da beklerim https://mylens.com.tr/ 🙂 İyi günler dilerimm..
Zina Ward raises some very insightful objections and I will try to address what I take to be the three main ones. Against the argument that cognitive constructs are adaptive and are therefore individuated externalistically, she points out that a cognitive system can have an adaptive function and yet be individuated by factors entirely “in the head”. After all, some neural systems, despite being internalistically individuated, have adaptive functions. I agree with Ward on this point: just because something has an adaptive function and has evolved to fulfill some purpose, we don’t have to individuate it etiologically or with reference to its adaptive history. So we can individuate airplane wings and bird wings both as wings, and lump them in the same category for some purposes, yet split them into two different categories if we’re interested in the way in which, say, bird wings evolved from more primitive structures. The reason that we individuate in terms of causal history (at least sometimes) with cognitive capacities has to do with the computational level of explanation and the importance of understanding what something has evolved for in order to understand what it is. That doesn’t mean that we couldn’t individuate cognitive capacities in terms of narrow function, but in cognitive inquiries, we often choose not to. It also doesn’t mean that we couldn’t individuate neural structures etiologically, since like all biological structures they are also under selection pressure. But the selection pressures are not identical in the two cases. The kinds of pressures that select for a capacity of episodic memory are rather different than those that select for, say, a hippocampus. So even when we investigate neural mechanisms etiologically, it is not the same causal history that is relevant to their individuation.
On a related issue, Ward claims that when it comes to episodic memory in particular, I admit that states of episodic memory need not be individuated etiologically, since I allow that “having a link to past experience is in fact not part of the individuation criteria for an episodic memory state.” Here, I should clarify that I distinguish the capacity of episodic memory from states of episodic memory, which are generated by it. States of episodic memory are, I claim, individuated etiologically in the ontogenetic sense, since (I propose that) what makes them such states is that they are produced by the capacity of episodic memory. But it is possible that a state produced by the capacity of episodic memory is not causally linked to episodes in the thinker’s past. That is just because our memory capacity sometimes appears to produce ersatz memories. Thus, what makes a state an episodic memory in the first place is that it is a product of the capacity of episodic memory, but what makes it a memory of, say, my tenth birthday is that it bears a trace of that event.
A third challenge that Ward poses has to do with the broader metaphysical issue of individuation. She points out that on occasion, I explicitly equate the identification and individuation of a kind, and that I seem “to conflate how a state is characterized with how it is individuated.” Similarly, she holds that “etiological considerations seem evidentially relevant to concept ascription, but they do not individuate the concept animal.” Let me start with a terminological point. It’s true that at times, I use “individuation” and “identification” interchangeably, and I should have taken more care to say that I was using “identification” in the sense of providing identity conditions, not in the sense of recognizing or diagnosing. What I have in mind by individuation (or identification, in this sense) of a kind is specifying the factors that make it the kind it is. I hold that many cognitive kinds are externalistically individuated. Ward says rightly that a (type of) cognitive state “is externalistically individuated when factors ‘outside’ of the thinker determine whether they are in that state or not. Holding fixed everything in the head, toggling facts about the environment or the thinker’s causal history changes whether the state is present or absent.” I would say that externalistic individuation is at play either when you hold fixed everything “in the head” and change the environment or causal history and the state changes, or when what’s “in the head” varies and the environment or causal history are the same and the state remains the same. When it comes to children and some concepts, I would claim that the latter is what’s at issue. Children have rather different discriminatory and inferential abilities than adults when it comes to reasoning about animals (what’s “in the head” is very different), yet we often attribute the concept ANIMAL to them because they are in tune with the same external environment. So what it is to possess the concept ANIMAL consists in part in having a certain causal history. This is not just a matter of “evidential relevance” but of what makes something an instance of the kind.
Thanks for the reply, Muhammad Ali! It sounds like we actually have quite a bit of common ground. One more thought about the episodic memory example: If an episodic memory state is etiologically individuated because it is produced by the capacity of episodic memory, then etiological individuation is a somewhat weaker notion than I originally took it to be. It is much easier to imagine neural states that are etiologically individuated in this sense – e.g., a particular oscillatory firing pattern might be “etiologically individuated” because its identity depends on having originated in a particular cortical nucleus. So I would worry that this notion of etiological individuation (on which etiology includes “head-internal” causal histories) would make it harder to establish the other half of the mismatch claim – i.e., that neural kinds are internalistically individuated – and thus create difficulties for the Individuation Argument.
Your comments on the concepts case gives me a lot to think about. I was particularly intrigued by the idea that there is externalistic individuation of a cognitive state “when what’s ‘in the head’ varies and the environment or causal history are the same and the state remains the same.” I suppose a cognitive state that is present IN VIRTUE OF the environment or causal history remaining the same is externalistically individuated. But it seems like a cognitive kind that is internalistically individuated could nevertheless be robust to a variety of in-the-head conditions.