Equipped with an adequate notion of conceptuality, we now proceed to show that concept-possession is exactly what is needed for a cognitive system to overcome the specific limitations associated with knowledge-how, and hence be able to gain access to the level of knowledge-that. Why exactly is it necessary for a system to possess concept-like representations in order to have knowledge-that as opposed to knowledge-how?
Context-bound versus context-free knowledge. For this polar contrast the answer, in short, will be that conceptual representations are precisely those representations which make the subject able to generalize information over a range of different behavioral situations. Conceptual representations are, as we have seen above, representations whose functional role is to classify aspects of a scene, or items of a behavioral pattern, according to a certain category. This is the reason why only conceptual knowledge (whether verbally expressible or not) can enable overcoming the limits of situationally-bound use. Intuitively sampling objects, for example, on the basis of some salient similarity criterion, is a manifestation of knowledge-how, because it depends on situational features—for instance that the situation represents some sort of average type to which the corresponding behavior is adapted. To overcome such situational limitations, categorical distinctions have to be introduced that enable the subject to transfer his or her knowledge partly to new situations that deviate, for instance, with respect to the objects that have been treated in standard situations. For example, a waiter who starts to work in a new restaurant using only coffee cups of one type, that is slightly higher than the large type used in the former restaurant, might fail in balancing the cup as long as he only takes recourse to his knowledge-how; but he might be more successful if he relied on a conceptual understanding of a distinguished large-cup-technique. In the same way, anticipation of the flight of a javelin is a situation-bound ability, since it depends on relatively rigid processing of visual information and proprioceptive mechanisms that are well-adapted to a range of standard cases, but fail for cases outside that range. If the case is exceptional (e.g., strong wind from behind), the subject can only attain success by analyzing the influence that this particular external condition will have on the standard performance. The same applies for knowledge-how expert chess knowledge, which fails in cases of random constellations because the experts’ expertise in evaluating the scene is dependent on average situational features. The occurrence of “new” constellations requires extracting general properties from the scene, and thus has to be done by means of conceptual representations.
Impenetrability versus penetrability of knowledge is a contrast almost built into the notion of conceptuality that we propose. Non-conceptual representations are non-receptive for additional stimuli that could yield classificatory behavior. They have to be non-receptive (“impenetrable”) in order to avoid interferences that could disrupt the more or less rigid mechanisms by which some well-defined type of behavior is regularly produced. Impenetrable knowledge-how, for example, is manifested by navigating ants calculating their way home according to some rigid computational processes that are deployed on the basis of a small number of parameters. If the experimenter interferes with the process by repositioning the ant, the mechanism still works as it would have done without relocation, with the result that the ant misses the nest by exactly the distance and direction to which it has been repositioned by the experimenter (see Bartels & May 2009). In contrast, conceptually-based processing has to be penetrable in order to guarantee that categorical information can be extracted from the scene according to specific stimuli (in this case the repositioning stimuli) and used in evaluating the result produced by rigid processing up to the time of repositioning.
Implicit versus explicit knowledge. This distinction refers to whether or not the knowing organism has knowledge of the rules governing its knowledge application. For example, people learn the grammar of their natural language or internalize their society’s norms implicitly, that is without knowledge of the principles that guide their language use or their social behavior. In such cases people represent rules only indirectly, by means of dispositions to have their reactions determined by the linguistic or social information in a way that can be recognized by their fellow subjects as to be in accordance with the rules. In contrast, explicit knowledge requires direct representation of rules, objects, or properties. The waiter in the restaurant, for example, after having achieved knowledge-that about his balancing of coffee cups, is able to refer directly to two sorts of cup shape, the high and cylindrical or the flat and bowl-shaped, respectively. In other words, he must be able to represent properties; if so, the waiter would, for instance, be able to draw inferences from the contents of his knowledge. Now, a person’s ability to produce attribute-representations of objects presupposes the ability to apply categories to his or her own experience. For example, the waiter is able to represent coffee cups as high and cylindrical objects because his capacities include the ability to apply the category of shape to the objects he is balancing. Thus, a person’s possession of conceptual capacities is a condition that has to be fulfilled for his or her knowledge to be explicit. Moreover, given that the additional conditions for conscious processing of cognitive representations are fulfilled, the subject would then be able to consciously think about and to draw conscious inferences about the objects. In addition, verbalizability of knowledge depends on the presence of this conscious form of explicit knowledge.
Automatic versus effortful processing. As we have argued in (B), conceptuality entails openness to penetration. Now, if cognitive processing is receptive to penetration, additional costs in terms of attention and additional processing necessarily occur. If the ant’s navigation mechanisms were receptive to a certain type of repositioning, it would have to use additional computational pathways for processing “repositioning information” and would be in need of additional calculation to determine the influence of the particular repositioning on the result produced by rigid calculation of the expected path back home.
Continuous versus discontinuous processing. Knowledge-that is characteristically used in a step-by-step manner with intermediate knowledge states (discontinuous), whereas knowledge-how appears to be grounded in smooth and fluent processing without intermediate states (continuous). The difference can be accounted for by the fact that knowledge-that is grounded in concept-based processing allowing for and instantiating discrete inferential steps, whereas knowledge-how is based on concept-free processing without clearly-defined intermediate knowledge states. An observable consequence of the continuous nature of knowledge-how is that lapses in knowledge use result in graded errors, or continuous distributions of errors (e.g., gradual precision losses of sensorimotor movements), while lapses in use of knowledge-that express themselves in categorical errors, or discontinuous error distributions (e.g., switches of categories or total failures to come up with a result).
It is beyond the scope of the present article to give an outline of a research agenda for empirically confirming and underpinning the present account of knowledge-how compared to knowledge-that. Different examples of potential research areas and experimental paradigms have been pointed out in the preceding sections (e.g., numerosity judgments, spatial memory, intuitive knowledge use). The most convincing way to support the adequacy of the conceptuality criterion for distinguishing between knowledge-how and knowledge-that will be to run new experiments in these or other research areas that reveal behavioral and/or neural dissociations that comply with the distinction between concept-driven vs. concept-free knowledge-use along the lines of the different peculiarities of practical knowledge outlined above.