Know-how, intellectualism, and memory systems

ABSTRACT

A longstanding tradition in philosophy distinguishes between knowthatand know-how. This traditional “anti-intellectualist” view is soentrenched in folk psychology that it is often invoked in supportof an allegedly equivalent distinction between explicit and implicitmemory, derived from the so-called “standard model of memory.”In the last two decades, the received philosophical view has beenchallenged by an “intellectualist” view of know-how. Surprisingly, defenders of the anti-intellectualist view have turned to the cognitivescience of memory, and to the standard model in particular, todefend their view. Here, I argue that this strategy is a mistake. As it turns out, upon closer scrutiny, the evidence from cognitivepsychology and neuroscience of memory does not support theanti-intellectualist approach, mainly because the standard modelof memory is likely wrong. However, this need not be interpretedas good news for the intellectualist, for it is not clear that theempirical evidence necessarily supports their view either. I arguethat, currently, the philosophical debate is couched in terms thatdo not correspond to categories in psychological science. As aresult, the debate has to either be re-interpreted in a vocabularythat is amenable to experimental scrutiny, or it cannot be settledempirically.

KEYWORDS:

• Knowing-how
• knowing-that
• intellectualism
• declarative memory
• H.M.
• procedural memory

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Stanley’s argument need not be read as two-tiered. One could read it as inferring – from the linguistic analysis and from the view that the semantics for know-how ascriptions delivers their truth conditions – that the truth makers of such ascriptions are cognitive states of propositional knowledge. This reconstruction is probably accurate too. My rendition simply wants to leave open the possibility that one’s commitments to the truth conditions of our know-how ascriptions need not carry ontological weight (I discuss this possibility for propositional attitudes in De Brigard, 2015).

2. To the best of my knowledge, the first time Squire published his now famous diagram, the main split occurred between declarative and procedural memory, which, in turn, he sub-divided into skills, priming, simple classical conditioning, and other (Squire, 1986). Two years later, and perhaps in an attempt to unify the vocabulary employed by other researchers that contributed to perfecting the model, procedural memory became a sub-class of non-declarative memory, covering motor, perceptual, and cognitive skills (Squire & Zola-Morgan, 1988). Figure 1 is based on Squire and Zola-Morgan’s 1991 diagram, in which procedural memory is now simply called memory for skills and habits.

3. The idea that there are different kinds of memory is not new. At least since Aristotle, both philosophers and psychologists have argued for different kinds of memory (De Brigard, 2014a; Michaelian & Sutton, 2017). Memory taxonomies that aimed to fit empirical evidence, however, were less common. In this regard, Squire’s SMM was not the only model available at that time. Tulving (1985), for instance, had suggested a different, nested model based upon single rather than double dissociations, whereby episodic memory depended (i.e., phylogenetically, ontogenetically, and causally) upon semantic memory, which, in turn, depended (i.e., phylogenetically, ontogenetically, and causally) upon procedural memory. Nevertheless, the SMM prevailed, despite its strong commitments to double dissociations, in part – I think – because it was better suited to fit non-human animal evidence while also being able to accomodate the evidence accounted for by Tulving’s model (for an opinionated historical review, see Squire, 2004).

4. Unlike perceptual priming, which is measured by the increased probability of responding with the same target item as the prime when perceptual features are manipulated (e.g., both seeing and responding “envelope” in a word-completion task like “e_v_l_p_”), conceptual priming refers to the increased probability of responding with the same target item as the prime when the manipulation is conceptual (e.g., responding “envelope” in a recognition test more readily after reading about mail than about food). By contrast, semantic priming is understood as the increased probability of responding not with the same item as the prime but with a semantically related one.

5. Other aspects of H.M.’s “textbook” neuropsychological profile have been questioned too. For instance, it has been suggested that H.M.’s post-operative language was affected (MacKay, Stewart, & Burke, 1998) and that his working memory may have been impaired too, since it was only minimally tested (Ranganath & Blumenfeld, 2005). However, I prefer not to discuss these two observations at length, mainly because the issue about H.M’s postmorbid linguistic abilities is very hard to settle (see Corkin, 2013, Ch., p. 11) and also because the evidence regarding his working memory capacity is too limited.

6. There are two other studies worth mentioning, one of which is the main topic of Section 5. The other one is a brief report by Yamashita from 1993, where three patients with bilateral hippocampal damage are compared against six controls in the rotary pursuit task. Unfortunately, this study is severely underpowered, and the variance in the data from both patients and controls is so large that it is almost impossible to interpret.

7. Transient global amnesia (TGA) is a rare neuropsychological disorder, usually caused by a temporary anoxic lesion to hippocampal neurons in CA1, and it is normally reversible. During the acute phase, which lasts about 24 hours, patients with TGA present a profound amnesic profile, after which they recover back to baseline but without remembering anything of what happened during that period.

8. Many researchers believe that episodic memories are retrieved by the combination of two relatively distinct sub-processes: recollection and familiarity. The former is characterized as the slow, intentional retrieval of the rich, contextual spatiotemporal information that constitutes the content of our episodic memories, whereas the latter is a fast, unintentional retrieval of the general gist of the event, which brings about the impression that it occurred in one’s own past. Moreover, there is now evidence of underlying neural distinctions between the two processes: While recollection has been associated with activity in the hippocampus proper and has been indexed by posterior late positive components using event-related potentials (ERP), familiarity has been associated with activity in the rhinal cortex and with anterior early negative ERP components (Skinner & Fernandes, 2007).

9. Roy and Park’s motivation to include the RC trial in S3 stems from the observation that, during S1 and S2, D.A. “made comments suggesting that he knew the function of the tool, but that he did not know how to position to recipient appropriately” (Roy & Park, 2010, p. 3031). Similar observations can be found in Corkin’s report of H.M.’s performance. She reports, for instance, that “at the begging of Rotary-Pursuit and Bimanual-Tracking test session he was allowed to look at the apparatus and then asked to describe the task. His memory for the Rotary-Pursuit task, though somewhat inaccurate, was consistent in specifying that he had to touch the stylus to the target in order to stop the disc from turning. On one occasion he further stated that he was not supposed to “touch that spring part” on the stylus, something that he had in fact been reminded about several times before. H.M.’s description of the Bimanual-Tracking task was consistently accurate from Session III on.” (Corkin, 1968, p. 264). Corking called the retention of this fragmentary information “testing habits,” and although she did not elaborate on the nature of this retention, it is consistent with her view that H.M.’s capacity to remember information about the task may have contributed to his performance, just as his deficits in remembering it may help to explain why his performance was never optimal.

10. I wanted to thank the audiences of the Workshop on Memory and Skill (Duke, 2016), the Instituto de Investigaciones Filosóficas (UNAM, 2016), and the departments of philosophy at the Universidad Nacional de Colombia (2017) and Tufts University (2018). Special thanks to Carl Craver, Jackie DeRosa, John Krakauer, Bryce Gessell, Paul Henne, Bryce Huebner, Kirk Michaelian, Carlotta Pavese, and Walter Sinnott-Armstrong for comments on previous drafts.