The paper reviewed here is ‘Social Concepts are Represented in the Superior Anterior Temporal Cortex’ by Zahn and colleagues and freely available here. The title of the paper is bold and suggests that social concepts are represented in a circumscribed part of the brain on the basis of the results of this study. They select 26 participants with an equal gender distribution. The sample group are not representative of the population averaging 29 years of age and having an average of 17.5 years of education suggesting that they have more years of education than the general population. Given that this is a relatively young population the exclusion of psychiatric and neurological illness as well as use of centrally acting medication means that they are less likely to be representative of the general population as a small percentage of the population in this age group would be expected to have depression and other illnesses. I wasn’t clear on why a measure of self-esteem was used. The paradigm involved presentation of fixed visual patterns, negative and positive social concepts and animal concepts. I think the visual patterns were included perhaps to control for activity related to presentation of visual information. However this may not be an effective control as it wouldn’t necessarily activate language areas in the brain in contrast with the word pairs. The authors have used word pairs derived from the work of other researchers with relevant citations included. There is a useful supplementary PDF which contains more information on the methodology. Subjects were scanned using a 3-Tesla scanner and acquiring T2-weighted images. The researchers then compare the activation patterns with different stimulus presentations.
One difficulty I have with the experimental paradigm is that the storage of social concepts is being equated with the response to the presentation of word-pairs. Why is the activation not just due to the presentation of words? One argument might be that if it was due to the presentation of words then the activation with animal words should produce a similar response. So if there is a difference, then it is most likely due to the content of the word-pairs. The researchers have gone to a lot of effort to obtain normalised data on the word-pairs so that they may have some confidence in the validity of the concepts being investigated. It is relatively straightforward to visualise the activation of a ‘concept’ by the words and that this concept would then be detected as activity in the region where it is stored. However this chain of reasoning contains a number of assumptions each of which can be challenged. Firstly does the presentation of a ‘social concept’ word immediately trigger the ‘concept’ in the reader. Perhaps ‘concept’ is a little too vague. For instance what is the concept of ‘loyal’ as described in this paper. For one person this may trigger a memory of an event, an event which they then associated with the concept. For another person it may be a person that they think of as loyal. It may be visualised as an action or a perception. Each type of association may be located in different brain regions and yet each is being labelled as the same ‘concept’. Thus I would argue that the term ‘concept’ is insufficiently defined and a standardised technique is required in which the same perceptual response is produced in the reader. This however is unlikely as each word evokes in different readers, different responses in accordance with their lived experiences. So the next question would be – why would they be the same? I find this a little dubious and suggest that aggregating the data from different subjects would not make sense as this would average out different activated regions. If these regions are activated differently then this means that useful information to the question being asked is being lost in the averaging process. Another interesting point is why is a concept static? I can think of the concept of a ball and as I focus on this, I can imagine it moving through the air or else changing shape or size and yet it remains the same concept – a ball. However within my brain, I would expect slightly different patterns of activation corresponding to these different ‘expressions’ of the same concept.
Instead, with the average process, I would argue that the best we can achieve is to say which area of the brain is most likely to be involved in the interpretation of the relevant word-pairs – which does not necessarily equate to ‘concept’. For instance it may be that the word-information needs to be passed to different areas to activate associations and the superior anterior temporal cortex is where that process begins, but the experiencing of the associated concept happens in these other areas. Using this argument it is possible to see that the nice voxel images seen in the paper can actually have a number of meanings depending on the phenomenological interpretation of the study design – and there are several such interpretations. So this is as much philosophy and phenomenology as it is about cerebral blood flow. But for argument’s sake let us suppose that there is an associated concept, that people have a single associated concept and that the averaging helps to precisely localise this area of activation – what other problems can there be?
Well, if this is indeed the concept associated with a presented word then we know that blood flow doesn’t necessarily correspond with the activation pattern – they can be separated in time because of the way that the neurons meet their energy requirements by regulating local cerebral blood flow. Even if they were identical though there is the additional problem that the information is not particularly well represented spatially on a 3 T scanner and even if it was then how do we know that we are looking at the right activation patterns? What happens if the associated concept is represented by a subtle shift in the firing pattern of the neurons which actually requires no change in cerebral blood flow. For instance, we know in neural network paradigms that the overall pattern of activity and the corresponding stored memory can change but there doesn’t need to be any change in the overall energy requirements of the network. It can also be argued that there is a selection bias in that regions of interest are selected on the basis of the literature. If these results are valid, then the same test reproduced on the same group should produce activation in the same region and a larger replication study should confirm these findings. Such replication would be important as the functions discussed here relate to semantic dementia and/or frontotemporal dementia.
If you have any comments, you can leave them below or alternatively e-mail firstname.lastname@example.org
The comments made here represent the opinions of the author and do not represent the profession or any body/organisation. The comments made here are not meant as a source of medical advice and those seeking medical advice are advised to consult with their own doctor. The author is not responsible for the contents of any external sites that are linked to in this blog.