An adequate meta-theory of human knowledge should be able to account for empirical differences observed when people use practical rather than theoretical knowledge in the most general terms, and be able to deliver an explanation for these differences. The starting point for the need to distinguish between practical and theoretical knowledge is the behavioral and neurological differences or dissociations in performance in different sensory, motor, or cognitive tasks, e.g., performance differences between experts and novices, between normal and prefrontal patients, between DF and IG. In actual research observed behavioral or neurological differences and dissociations are often accounted for by describing them in terms of polar opposite knowledge attributes or effects. In our understanding this is a first step in the direction of a theory of knowledge-how, even if it is still short of delivering a satisfactory explanation of the observed behavioral and neurological differences.
In the cognitive science and psychological literature, one finds the following polar opposite ascriptions of attributes of knowledge-how as opposed to knowledge-that:
A. Context-bound versus context-free knowledge. Knowledge-how is specific to the domain or the situation of its use, whereas knowledge-that is not. In other words, knowledge-how is about situational skills, while knowledge-that is about general facts (e.g., Clark 1997; Clancey 1997). For example, throwing a javelin and anticipating its movement when it leaves the hand is a case of context-bound knowledge, whereas calculating the biomechanical forces needed for optimal performances (e.g., the ballistics of an optimal flight trajectory) is an instance of context-free knowledge. Chess experts as compared to novices have superior context-bound knowledge of constellations of chess figures, which helps them to reproduce specific shortly-presented board situations from memory. However, their superior knowledge does not help expert chess players to reproduce random constellations of chess figures from memory, as their skill for applying context-bound perceptual chunking mechanisms on meaningful constellations of figures does not prove beneficial.
B. Impenetrability versus penetrability of knowledge. Knowledge-that is penetrable by other cognitive processes or meta-processing, whereas knowledge-how is impenetrable (Pylyshyn 1984, 1990). Impenetrability means that use of knowledge-how is not changed by internally (e.g., beliefs, goals) or externally (e.g., distracting stimuli) triggered cognitive processes. One example is subitizing, i.e., the rapid, accurate, and confident estimation of the number of displayed elements (e.g., stones), which works fine and is robust against internal or external distractions. In contrast, the use of knowledge-that to determine the number of regularly arranged objects by counting them or doing mental arithmetic (e.g., adding over rows of elements 3+5+4+2+…) is prone to interferences from internally- or externally-activated cognitive processes. If athletes change the order of different sensorimotor sub-processes (e.g., in technical sport disciplines such as high-jumping or hitting a golf ball), they can encounter considerable problems and might need additional time and effort to build up new knowledge-how. Not so well-trained movements (e.g., dancing steps in beginners) can be more easily rearranged.
C. Implicit versus explicit knowledge. Use of knowledge-how takes place largely outside of awareness and hence cannot be verbalized, while knowledge-that is to a large degree consciously available and can be verbalized. In the last decades psychological research has made substantial progress in distinguishing between implicit and explicit forms of human learning, memory, and information processing (e.g., Dijksterhuis & Nordgren 2006). People learn the grammar of natural language or internalize their society’s norms implicitly, that is, without conscious knowledge of the principles that guide their language use or their social behavior (e.g., Reber 1989). Implicit memory is, for example, displayed in cases of amnesia, in which patients are not able to explicitly recall previously-presented items or events from memory, while performances on tasks that do not require explicit memory such as perceptual priming or sensorimotor skills are undisturbed and virtually normal (e.g., Tulving & Schacter 1990).
D. Automatic versus effortful processing. Use of knowledge-how is automatic in the sense that it requires little attentional monitoring or guidance, and in the sense that that its demands on working memory are quite low (Bargh & Chartrand 1999). Use of knowledge-that is generally more effortful, and can be shown to require significant attentional as well as working memory resources (Hasher & Zacks 1979). Good examples of the distinction between the automatic and effortful use of knowledge can be found in the domain of spatial cognition: Blindfolded navigators (animals as well as humans) complete triangles by returning to the starting point on the basis of automatic vestibular and kinesthetic path-integration mechanisms (knowledge-how), while only humans are able to use effortful geometrical calculations (knowledge-that) to find their way back to the origin of the outbound travel. Experiments show that simultaneous secondary tasks (e.g., to-be-ignored spatial movements vs. counting operations) differentially affect the one or the other type of knowledge processing (May & Klatzky 2000). To give another example from research on spatial cognition: Wayfinding on the basis of multimodal sensory inputs from the surroundings and from automatic updating is very different from the quite effortful and highly disturbable use of knowledge-that that results from listening to verbal route-descriptions or maps (Montello 2005).
E. Continuous versus discontinuous processing. Use of knowledge-how expresses itself in smooth and continuous processing, while knowledge-that is normally reflected in step-by-step processing along a discontinuous path of intermediate knowledge states. Recent dynamic systems accounts of the sensory, motor, and cognitive processes underlying human knowledge use describe these differences in terms of different attractor landscapes of mental or neural state spaces (Spivey 2008). Research into children’s cognitive development, for example, reveals that there are two levels of spatial location coding in memory. In a first phase, children learn to code the metric distance between locations (e.g., allowing them to find previously hidden objects in terms of distance from the sides or the corners of a rectangular sandbox). In a second phase, children attain the ability to impose organization on their spatial knowledge (e.g., allowing them to divide the spatial layout in hierarchical subsections or regions). The shift from the first to the second level reveals itself in changes in the types of spatial errors (discontinuous vs. continuous distributions) children commit when locating hidden objects (Newcombe & Huttenlocher 2000).
This list of opposing attribute pairs is probably not complete, but seems a good starting point for our purposes. It can be thought of as a general description and characterization of practical knowledge in contrast to theoretical knowledge. Not every single case of knowledge use will be easily describable by means of the list, or will even require a full description along all opposing attribute pairs. However, chances are good that the overwhelming majority of cases will be adequately described by using such a set of opposing attributes, and, generally, the profile over the five attributes will correctly apply. We will argue that this list of attribute pairs, together with their predominant assignment to the one or other knowledge variety, is what an adequate and fruitful theory of knowledge-how vs. knowledge-that should be able to account for.