In a previous post (here) I remarked on a pair of papers that I had not be able to read in full but only had the abstracts. A kind reader, G. Marchetti (http://www.mind-consciousness-language.com), has let me see these papers (citations below). I am relieved that I did not make any ‘oopses’ in understanding the abstracts. There were ideas in the papers that I didn’t touch on in that post and will now.
First the Bilalic paper:
Recognizing human faces is one of the most essential visual skills—and also one of the most practiced ones. Since the very beginning of our lives, we have been exposed to faces as a major source of social information. The neural substrates of face recognition have been extensively studied. One of the most important brain structures for face perception is the fusiform face area (FFA), located in the right lateral part of the midfusiform gyrus. Some researchers even proposed that the FFA is a specific module exclusively devoted to face recognition. This face-specificity hypothesis contrasts with the expertise hypothesis, which maintains the FFA is a general expertise module specialized for perceptual processes associated with visual individuation.
This idea of individuation is important here. We do not view faces as types as much as we think of them as particular people, often with their own name. This is an area for the kind of discrimination that carries a proper noun for identification. Presumably this is the area that would deal with any sort of object where a large number of similar ones must be treated as each being one of a kind. Off hand, chess does not seem ideal as a source of individuation. But it does seem a good source of similar patterns with very different significances and easy to experiment with (starting with an expert rating available).
They showed that faces activate the FFA more than chess boards – this was no surprise as the subjects see more faces than boards. Also they showed that the activation of chess boards was expertise-modulated, and therefore being used by chess experts.
It is difficult to explain our results solely with attentional effects. IPS, an attention-related area, was engaged in all tasks, but there were no differences between experts and novices. … indicates that the FFA activation was probably independent of task difficulty and the attentional processes necessary in this particular context.
The FFA is not distinguishing just patterns, but meaningful ones. It seems that pattern is not enough and the ‘intent’ component of real games is need. I think this is very important. Chess boards are very unlike faces and other types of expertise that have been studied in the FFA (for example bird watchers). But chess patterns are meaningful; they can be view in a holistic way by experts and they have some of the spirit of intentions/goals/rules/meaning.
An additional piece of evidence that the FFA effects are not related to the mere complexity is the pattern of results on normal and random positions. Both position types are comparable in that they involve a similar number of chess objects forming interrelations on the same full chess board. Only normal positions, however, contain relational patterns between chess pieces that are meaningful to experts. FFA appears to be responsive to this subtle distinction, as shown by the different activation levels between normal and random positions among experts only.
The non-visual and the non-holistic processing of chess boards and of faces takes place elsewhere:
Our results indicate that FFA is not directly related to core expertise processes but that it may support some of them indirectly by processing the stimuli holistically. The real utilization of stored chess knowledge by experts seems to be mediated by the collateral sulcus. It should be noted, though, that even in face perception, we have a dedicated network of brain structures, which are responsible for different processes. … Simple stimuli (e.g., isolated chess pieces), which do not consist of complex relational patterns formed by clearly distinct individual elements, may not engage holistic processing properties of the FFA. In contrast, naturalistic multipart stimuli (e.g., faces, full-board chess positions) seem to invite holistic processing in experts, mediated by increased FFA activity.
The Bilalic paper used real people with real FFAs but a expertise that is not typical. The Tong paper on the other hand uses more typical classes of objects but computer models of the learning process instead of people. It is also an older paper.
They start with a interesting definition of what an expert is rather than the ready made ranking of Bilalic’s chess players.
We use Gauthier’s operational definition of the term: experts are as fast to verify that a picture of an object is a particular individual (subordinate level) as they are to verify their category membership (basic level). For example, a bird expert would be as fast and as accurate at verifying that a picture of a bird is an “Indigo Bunting” as at identifying it as a “bird.” On the other hand, a novice will show the fastest reaction time at the basic level, and is slower at both subordinate and superordinate level. The basic level was first identified by Rosch as the level at which objects tend to share the same shape and function, and tend to correspond to the first word we use to describe an object (a picture of a chair is labeled “chair” rather than “furniture” or “office chair”). When training a subject in a novel category, the downward shift in reaction times in this task is taken as evidence of expertise.
The experiments were done with electronic networks which ‘learn’ to make distinctions by trial and error. The networks change the weights of the signals between ‘neurons’ in the hidden layers between input and output when they are learning. The hidden layers can then be characterized by the researchers. The researchers used two types of input/output which made the networks learn in the style of non-experts (basic) and of experts.
By analyzing the hidden layers of the two types of networks, we found that expert networks spread out the representations of similar objects in order to distinguish them. Conversely, basic networks represent invariances among category members, and hence compress them into a small region of representational space. The transformation performed by expert networks (i.e., magnifying differences) generalizes to new categories, leading to faster learning. The simulations predict that FFA neurons will have highly variable responses across members of an expert category.
What does this tell us to expect in the FFA?
What the results do suggest is that if the FFA is performing fine-level discrimination, then that task requires it to develop representations of the stimuli that separate them in representational space—the neural responses are highly differentiated. That is, similar objects have the differences between them magnified by the expert networks. On the other hand, networks that simply categorize objects map those objects into small, localized regions in representation space (this is in the space of neural firing patterns, and should not be confused with spatially localized representations).
What does this say about consciousness? This blog is, after all, about consciousness. There is a saying – if you don’t know how it’s done then it’s easy. Recognizing individuals by their faces, voices, whole bodies etc. just effortlessly happens for us – it just happens and therefore it seems easy. If it were a conscious process it would not be invisible, we would know how it is done (or at least the steps involved) and it would be seem hard. To us recognizing people and even predicting their actions ‘just happens’ and anything that ‘just happens’ involves an enormous amount of processing that is transparent to consciousness. Pattern recognition is not a conscious thing – by its nature it cannot be.
Tong, M., Joyce, C., & Cottrell, G. (2008). Why is the fusiform face area recruited for novel categories of expertise? A neurocomputational investigation Brain Research, 1202, 14-24 DOI: 10.1016/j.brainres.2007.06.079
Bilalić M, Langner R, Ulrich R, & Grodd W (2011). Many faces of expertise: fusiform face area in chess experts and novices. The Journal of neuroscience : the official journal of the Society for Neuroscience, 31 (28), 10206-14 PMID: 21752997