Intellectualism and the argument from cognitive science

ABSTRACT

Intellectualism is the claim that practical knowledge or ‘know-how’ is a kind of propositional knowledge. The debate over intellectualism has appealed to two different kinds of evidence, semantic and scientific. This paper concerns the relationship between intellectualist arguments based on truth-conditional semantics of practical knowledge ascriptions and anti-intellectualist arguments based on cognitive science and propositional representation. The first half of the paper argues that the anti-intellectualist argument from cognitive science rests on a naturalistic approach to metaphysics: Its proponents assume that findings from cognitive science provide evidence about the nature of mental states. We demonstrate that this fact has been overlooked in the ensuing debate, resulting in inconsistency and confusion. Defenders of the semantic approach to intellectualism engage with the argument from cognitive science in a way that implicitly endorses this naturalistic metaphysics, and they even rely on it to claim that cognitive science supports intellectualism. In the course of their arguments, however, they also reject that scientific findings can have metaphysical import. We argue that this situation is preventing productive debate about intellectualism, which would benefit from both sides being more transparent about their metaphilosophical assumptions.

KEYWORDS:

• Intellectualism
• knowledge-how
• knowledge-that
• procedural memory
• skill
• naturalism

Acknowledgments

We are grateful to Carl Craver for his feedback on an earlier version of this paper, to Jonathan Cohen for discussion, and to two anonymous reviewers for their helpful comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. According to dispositionalist views, for example, to stand in a relation to a proposition is to possess the appropriate disposition toward the proposition. (Believing that p might be a matter of being disposed to assent to p in the relevant circumstances, to use p as a premise in reasoning, to use p as a ground for doubting propositions that one knows entail ~p, and so on). On functionalist proposals, standing in a relation to a proposition is a matter of being in a state that plays a relevant causal role. (Believing that p might be the causal state that follows perceiving that p.) Representationalism proposes that standing in a relation to a proposition is a matter of tokening an internal representation with the same content as the proposition. It is also possible to hold these views in concert. For details about these views of propositional attitudes and their interrelations, see Schwitzgebel (2015).

2. On the other hand, there is evidence that ascriptions of practical knowledge do function like propositional attitude ascriptions, in the sense that they are non-extensional: Substituting co-referential terms in the content can alter the truth-value of the ascription. It might be true to say that a dancer knows how to give a performance of Improvisation No. 15 but false to say that the dancer knows how to give a semaphore recital of Gray’s ‘Elegy,’ despite the fact that the dance performed is a movement-perfect semaphore version of the poem (Carr, 1979).

3. By distinguishing between the semantic and scientific approaches to intellectualism, we are not committed to the claim that semantic theories and scientific theories are mutually exclusive. We are proposing, however, that the semantic theories to which Stanley and Williamson (2001) and Stanley (2011a, b) appeal in support of intellectualism are not scientific theories. They are theories which attempt to account for people’s semantic intuitions in terms of abstract entities like propositions, sets of worlds, and meanings, and thus they cannot offer a causal scientific explanation of how the intuitions arise. Following Fodor and Lepore (2012), we allow that semantic theories can be scientific theories when they are psychologized: when the semantic properties of linguistic expressions are mentally represented by competent speakers and when these concrete mental representations are causally responsible for those speakers’ semantic intuitions. As we make clear in this paper, neither Stanley and Williamson (2001) nor Stanley (2011a, b) has a psychologized causal theory of semantic intuitions: They do not think that propositions, for example, need to be understood representationally. In this paper, therefore, we can contrast their semantic approach with scientific approaches to intellectualism.

4. We use the term ‘cognitive science’ broadly here, to include all scientific approaches to the mind: We include implementation-level theories from neuroscience in addition to abstract computational theories, for example.

5. Fodor argues that cognitive science supports “intellectualist accounts of mental competences” (Fodor, 1968, p. 627), based on the claim that mental processes are computational processes. When a computer has a certain competence, such as knowing how to multiply, this competence can consist in running a sequence of rule-governed operations. If minds are computational, then our own mental competences might also consist in such sequences, even if we are unable to articulate the operations in question. Fodor proposes that it is therefore plausible that our practical knowledge or ‘know-how’ employs “propositions, maxims, or instructions” that constitute a formulation of our tacit knowledge (Fodor, 1968, p. 638).

6. The procedure that HM underwent in 1953 was intended to relieve the debilitating symptoms of epilepsy. It involved bilateral removal of the hippocampus, parahippocampal gyrus, entorhinal cortex, and most of the amygdalae. After the procedure, HM was unable to form new long-term memories of facts or events, (complete anterograde amnesia), and he could no longer access memories he acquired in the few years leading up to his surgery (retrograde amnesia). By 1968, however, Suzanne Corkin reported “the retrograde amnesia [of HM] is now restricted mainly to the year before his operation” (Corkin, 1968, p. 255).

7. The source of the declarative-procedural distinction is Cohen and Squire (1980). The more recent distinction between declarative and non-declarative memory stems from Squire and Zola-Morgan (1988). In characterizing it, the authors write “declarative memory includes what can be declared or brought to mind as a proposition or an image. […] Non-declarative memory refers to a heterogenous collection of abilities: motor skills, perceptual skills, and cognitive skills (these abilities and perhaps others are examples of procedural memory), as well as simple classical conditioning, adaptation-level effects, priming, and other instances in which experience alters performance without providing a basis for the conscious recollection of past events” (171).

8. It is a matter of debate what the relationship is between declarative memory and declarative knowledge. If not all declarative memories are factive (see e.g. Schwartz, 2018), and all declarative knowledge is factive, then some declarative memories are not declarative knowledge. Nevertheless, because it has been commonly assumed that all declarative knowledge is declarative memory and all procedural knowledge is procedural memory, and because declarative memory is dissociated from procedural memory, it follows that declarative knowledge is dissociated from procedural knowledge. This is the dissociation that the present argument from cognitive science rests upon.

9. As Fodor writes, “I don’t see how an organism can stand in an (interesting epistemic) relation to a proposition except by standing in a (causal/functional) relation to some token of a formula which expresses the proposition. […] I want a mechanism for the relation between organism and propositions, and the only one I can think of is mediation by internal representation.” (Fodor, 1978, p. 520) .

10. The following definitions of ‘propositional representation’ can be found in cognitive science textbooks: “Propositional representations [are] representations in which relationships are represented by symbols, as when the words of a language are used to represent objects and the relations between them” (Goldstein, 2011, p. 404); “According to the propositional hypothesis, mental representations take the form of abstract sentence-like structures” (Friedenberg & Silverman, 2006, p. 8).

11. Stanley proposes that “an anti-representationist view of the mental is completely consistent with the fact that we have propositional knowledge states” (Stanley, 2011a, p. 159), for example, and such a view makes clear that Stanley and Williamson's (2001) distinction between propositional and non-propositional mental states is intended “to be entirely neutral on how propositional knowledge is realized in the brain” (Stanley, 2011b, p. 45).

12. Metaphysical naturalism, also known as scientific metaphysics, comes in a variety of forms. For the purposes of this paper, we understand it as committed to the positive claim that scientific findings can be relevant to metaphysics, in the sense that our metaphysical conclusions can be informed by empirical scientific data. (For defenses of naturalized metaphysics, see Chakravartty, 2017; Ladyman & Ross, 2007; Maddy, 2007) We are not assuming that metaphysical naturalism must completely reject the role of a priori intuitions or conceptual analysis, or the evidential status of semantics (although several proponents of the argument from cognitive science do seem to be committed to this). Furthermore, we are not assuming that proponents of the semantic approach to intellectualism must be opposed to metaphysical naturalism; they might be pluralists in this sense. We are merely claiming that the argument from cognitive science assumes that findings from cognitive science can substantially inform our metaphysics of mental states (i.e., not merely by providing counterexamples to purportedly necessary truths).

13. The idea that our folk concepts are subject to change upon consideration of scientific evidence is key to naturalistic approaches to the mind, as Churchland argues: “What we take to be obvious or observational within the framework of intuitive psychology is not guaranteed correctness or survivability simply on the strength of that obviousness and observationality. […] even intuitive frameworks, and even observational concepts, can be reconfigured as science proceeds” (Churchland, 1986, p. 293).

14. The question of how practical meanings could appear as components of thoughts is one Pavese leaves “to future work” (Pavese, 2017, p. 94).

15. Fodor emphasizes that, for cognitive science, having a non-naturalistic theory of mental content is akin to having no theory of content (Fodor, 2008, p. 51). He proposes that “the referential kind of content affords by far the best hope of naturalization” (Fodor, 2008, p. 51); Rupert similarly acknowledges that referential theories of mental content “better satisfy naturalistic scruples” (Rupert, 2011, p. 102).

16. For further discussion of representational vehicles as non-semantic modes of presentation, see Aydede (2000), Schneider (2009), Drayson (2018).

17. In cognitive science, semantic compositionality plays a causal-explanatory role in virtue of representational (e.g. syntactic) compositionality. Pavese’s view would thus seem to require additional evidence to support the idea that we have a cognitive architecture which allows motor representations to be combined with amodal conceptual representations to form propositionally structured representations.

18. These constraints and rules include the isochrony principle (the tangential velocity of movements is scaled to their amplitude), Fitt’s law (the time required to move to a target area is a function of the distance to the target and its width), and the two-third power laws between curvature and velocity. In target-pointing experiments, subjects can adjust their pointing trajectories when the target is displaced without awareness of the displacement or their adjustments. See Mylopoulos and Pacherie (2017) for details.

19. There is another reason to question whether motor representations can account for Pavese’s practical modes of presentation. Pavese (2015) claims her approach to intellectualism is concerned solely with the mental properties which we ascribe to persons: knowledge, competence, and intentional action, for example. She explicitly contrasts this approach with the standard methodology of cognitive science as exemplified by Fodor (1968) – a method consisting in explaining the capacities of persons in terms of the properties and interactions of their ‘subpersonal’ cognitive subsystems: “I am interested in explaining competences that the subject can exercise intentionally. Because of that, the level of explanation at which my arguments work is personal [..] My view leaves it entirely open whether a Fodorian theory is the most adequate sub-personal theory of this sort of competences” (Pavese, 2015, p. 18). While there are a number of different ways to understand the personal-subpersonal distinction (see Drayson, 2012; Rupert, 2018), none of them would countenance Pavese’s (2017) appeal to the motor control system as personal-level or neutral with respect to subpersonal theories. The motor control system is a computational subsystem posited by cognitive scientists in order to give a decompositional explanation of the system-level capacities of the cognitive system – a mechanistic explanation of capacities like knowledge, competence, and intentional action, given in terms of the subpersonal states and processes that interact to produce these system-level features. Any attempt to explain the intentionally-exercised competences of a person in terms of the cognitive science of the motor system is no less subpersonal than Fodor’s (1968) theory. We charitably suggest that Pavese’s position has changed between the 2015 and 2017 papers.

20. This principled method might merely be inference to the best explanation. Williamson (2016) argues that when we construct philosophical theories, our evidence base is the total sum of human knowledge: scientific and semantic, a posteriori and a priori. Take note that, if any of the intellectualists discussed in this paper are using this strategy, they are not doing so in a principled way. They have not outlined the set of possible explanations nor the respects in which their own explanation is preferable.

21. Perhaps these philosophers are under the impression that their philosophical reflections, rather than empirical investigation, determine the explananda of cognitive science. As Rupert (2018) points out, some philosophers hold the view that their non-empirical characterizations of the mind “provide a presumptive constraint on cognitive-scientific modeling and theorizing” (Rupert, 2018, p. 6) in a way quite unlike that of any other science. Stanley (2011a, p. 147) gestures toward such a view when he suggests that folk psychological concepts constrain cognitive science in a way that folk concepts of physics don’t constrain the physical sciences. This is a view denounced by those working in cognitive science as well as by philosophers of science. See Thagard (2009) and Rupert (2018) for further discussion.

22. Weatherson observes that the focus on skill instead of know-how is “a trend that we see exemplified in recent work by, among others, Pavese (2013), Fridland (2014), and Tsai (2014)” (Weatherson, 2017, p. 371).

Additional information

Notes on contributors

Arieh Schwartz

Arieh Schwartz is a PhD candidate in Philosophy at the University of California, Davis. His primary research interest is naturalistic philosophy of memory.

Zoe Drayson

Zoe Drayson is Assistant Professor of Philosophy at the University of California, Davis. Her research focuses on issues at the intersection of philosophy and the mind sciences.