A cognitive account of skilled expertise

In a previous post, I raised three puzzles for our understanding of skilled epistemic practices - domains of skill where we attempt to acquire knowledge of the world, such as birding, archaeological surveying, art appraisal. These and other skills share a peculiar phenomenology, cross-cultural variability, and seem to confer prima facie justification. My aim in this and the next posts of this series is to provide a cognitive account of skilled expertise. I will look at how they are mastered, and at their position within a broader evolutionary framework.

Humans engage in epistemic actions—actions specifically deliberately aimed at gaining information, such as driving around in a new neighborhood to see the surroundings, putting coins in piles before counting them. Skilled epistemic practices are collections of epistemic actions, performed in a script-like, standardized way, to gain information. 

We are equipped with several cognitive tools to help us accomplish epistemic actions including ordinary perceptual capacities and early-developed domains of core cognition such as intuitive psychology and intuitive physics. I have argued in previous work that these ordinary capacities are crucial for the development of complex cultural practices, including the species concept in paleoanthropology and argumentation in natural theology. However, lots of the skills we have go well beyond these early-developed, "maturationally natural" domains of cognition (this term is derived from McCauley and avoids some of the problems of the concept of innateness - what is meant is domains of cognition that emerge early in development, with little teaching or other cultural prompting.

We acquire practiced skills mainly by learning them from others. Our special skills of discernment come about thanks to a host of cognitive adaptations for social learning and pedagogy, which are uniquely human. As Reid already speculated in his account of testimony, humans are uniquely trustful. Six-year-olds, for instance, are confident oxygen and germs exist on the basis of their parents' say-so, things they have never personally observed.

Interestingly, humans not only trust factual information but also that their teachers show them the best way to do things, trust in knowing-how. This is dramatically illustrated by series of experiments where chimpanzees and children both learned an elaborate way to open a puzzle box, but only children persisted in the hard way once it became obvious (with a transparent puzzle box) that there was a simpler way to open the box. This overimitation is not just present in western children (who might be influenced by western forms of schooling in their behavior), but has also been found in children from South-African and Australian hunter-gatherer cultures. Overimitation is not just a trait in young children (who are commonly seen as gullible), it actually increases with age. 

When we learn a skill from someone, we not only have to trust them the one time they tell us how to do something, but over and over when they tell us we are not doing it exactly right, or how we can improve. Think about the first karate kid movie, where Daniel has to put up with repeating the same movements over and over again under Mr Miyagi’s directions, which don’t make sense to him at the time, until he finally gets what they are for during a kumite (fighting for points).

Eastern training in martial arts is focused on this idea of getting the details of the skill right, such as in the so-called katas in karate. This requires a high degree of trust. Whereas experiments like Horner and Whiten’s are often interpreted as overimitation, one could argue that it was a critical element of the success of past human cultures that humans were willing and able to learn skills that are opaque to the learner.

How does the acquisition of skill happen? Skills rely both on environmental facilitators and on internal neural plasticity. Typically, skilled practices require some form of material culture to facilitate their transmission. Some of these material scaffolds continued to be used also by experts, such as books and articles by western scholars, or counting rods by ancient Chinese mathematicians. Others are used mainly during the learning process and are discarded once one has mastered the skill. 

Skills also take advantage of neural plasticity. As is well known in neuroscience, a repeated and persistent excitement of one neuron by another results in metabolic changes in both cells which increases their connectivity, a process known as long-term synaptic potentiation. This is how habits and routines, such as switching a light on before going to bed, are formed. For a long time, cognitive scientists such as Dreyfus thought that skilled actions are like routines. But now we know they are very different—this has been shown in diverse studies of sports and music performance, which show that people are sensitive to context and have a great deal of control over their skill, unlike in purely routine tasks (see here for discussion).

What explains the higher degree of control in skilled practices? The reason for this lies in pedagogy. Whereas routines are formed by everyday habits, the successful mastery of a skilled epistemic practice lies in deliberate practice, the practice skilled experts engage in to become well-versed in their domain, often centered on weaknesses. In the early 1990s, cognitive scientists thought deliberate practice was the only significant variable in the acquisition of skills. Now we know it is not the only thing, and that good teachers and a supportive social environment also play a role. In chess and music, the number of hours spent in deliberate practice explains about 30% of the variation in expertise.  Through deliberate practice–practice aimed at improving performance—we change our brain connectivity, and we do this in a different way than in pure routine-formation.

Consistently, fMRI studies found that acquiring any type of skill is associated with a reduction in prefrontal, medial frontal (such as anterior cingulate), posterior parietal, occipito-temporal, and cerebellar areas, a network that is consistently involved in attentional control. For instance, as adolescents become better at solving algebraic equations, there is a decrease in activity their prefrontal cortex  showing they need less executive control to complete the task, and additionally, they also showed a decrease in the left parietal region which is normally involved in spatial processing and number, indicating they needed to rely less on this as they got more fluent with using the algebraic equations.  There are less consistent increases in brain activation – what seems to happen is that when brain activity increases over time when acquiring an expert skill, it happens mostly in modular areas of motor and perceptual processing.  

People recruit specialized neural systems that have older phylogenetic functions in culturally novel domains of expertise. For example, the fusiform face area and occipital face area are involved in face recognition. Car, bird and dog experts recruit these areas to distinguish between Pontiacs and Oldsmobiles, or between collies and borzoi, presumably because like faces, they are members of a same category with visual features that differ in small details.  Gauthier et al. find support for this hypothesis by finding that people who are trained to visually discriminate between a novel class of objects that differ in details (greebles) also neural signature of face recognition. 

In sum, my cognitive account of skilled expertise proposes that cognitive adaptations for pedagogy and some extent of neural plasticity can explain how people can become experts in given cultural domains of skill. In the next post, I will explore how this can shed light on the peculiar phenomenology of these skills. 

Comments

[Elisa Freschi]

Helen, forgive me if I try to understand (and probably misrepresent something you said).

1. You say that skilled practice is significantly different than routine, insofar as skilled practice are context-sensitive, whereas routines are not.
--»Since you discuss the fMRI data concerning the acquisition of skilled practice, I wonder whether there is a significant difference in the acquisition of routines (I would expect this to be the case).

2. While you discuss the impact of deliberate practice, you state that the idea that deliberate practice alone would be enough (independently of teachers, I surmise) was current in the 1990s and has now been abandoned in favour of a more generous assessment of teachers and supportive environment, right? (I am asking because in connection with "30%" you link to too many articles on google scholars and I will not be able to read them all).

3. Last, I am not sure I can connect the paragraphs on deliberate practice and skilled practice with the last one on the increased ability to perceive minor differences. Do you mean to say that people learn through deliberate practice to reinvest brain areas into new tasks, so that they end up *perceiving* things, but only through the deliberate practice of learning how to perceive them?

thanks again for the stimulating mix of fMRI data and epistemology!

[Helen De Cruz]

1. Dear Elisa: Thank you for these comments! Yes, there is a significant difference in the acquisition of routines. Routines do not require deliberate practice- the thoughtful and deliberate way of practicing those elements of the skill you have not yet mastered. Take, for example, the light on the stairs. When I go up the stairs at night, I flip on the light switch - I do so because there is an association between [switch on light] and [go upstairs] when the condition [it's dark] is fulfilled. However, my lamp is broken and I have not yet replaced it. Still, I flip the switch without thinking because the association is there. My behavior is insensitive to the environment. By contrast, a skilled activity, though fluent, is responsive to all sorts of environmental factors, because I've trained it to be so. For instance, suppose I were an archer, I'd have practiced in wind-free conditions, strong wind conditions, hot and cold, light and dark, and thanks to the deliberate practice and neural wiring that comes with it, my archery would be responsive to all these conditions.

2. Correct - cognitive scientists thought deliberate practice was the only variable that mattered in acquiring an expert skill. Now we know that teachers, social environment, personal interest also matter. A rigorous study looking at these factors in chess and music can be found here: http://www.castonline.ilstu.edu/smith/405/readings_pdf/hambrick_deliberate_practice_2013.pdf

3. Yes - take the car and bird experts. Non-specialists use the fusiform face area as a visual area to discriminate between faces (broadly similar things that differ in details), but this network can be trained to become responsive to other categories of objects that differ in details. I've been told that grills of cars are really important to tell old models apart - a bit like the shape of eyes in faces. But it takes time and practice to get attuned to novel categories. First, you need to be on the lookout for the details that matter and seek them out consciously (e.g., headlights, grills), after a while with practice, your fusiform face are will pick them out and you need not use conscious control anymore to do it.