The core argument of the paper is that neural reuse is a central organising principle of the brain – in that it is common for neural circuits which have been established for one purpose to be cannibalised during both evolution and development and put to a new function without losing the older function. Michael Anderson analyses three other different theoretical approaches to reuse (the neural exploitation hypothesis, the shared circuits model, and neuronal recycling) and promotes his own version – massive redeployment – as superior. It makes the following three claims:
- Brain regions should support multiple cognitive functions in diverse task categories
- There is a correlation between the phylogenetic age of brain regions and frequency of deployment (the older the region the more it is redeployed)
- There is a correlation of a phylogenetic age of cognitive functions and the degree of localisation (the older the function the more it is anatomically localised)
A major part of our discussion related to these three claims. In particular, many wondered whether there were original weak modules that are evolutionarily old at the basis of the framework. And there was also a consideration about whether this is Lamarckian or involves the Baldwin effect.
In favour of his position he reviews the huge amount of research – mostly related to embodied cognition – which examines how perceptuomotor domains are redeployed in a range of other domains. For example, how spatial schemas are redeployed in mathematical capacities (what are now called SNAs – spatial numerical associations). Anderson also argues that although this evidence was originally proposed in the context of conceptual metaphor theory and concept empiricism, reanalysis of this evidence suggests that it outstrips these positions and points towards the need for a more general framework of neural reuse.
In addition to reviewing previous evidence, Anderson and colleagues also develop an amazing database – NICAM (Neuro-Image based Co-Activation Matrix) – which reviews 2603 studies in 824 journal articles. This gave them 21553 post-subtraction activations to be used to see to what extent specific brain regions support multiple functions. These can be subdivided as follows: 998 regions of interest
- Three action domains: execution; inhibition; observation
- Two perceptual domains: vision; audition
- Six “cognitive” domains: attention; emotion; language; mathematics; memory; reasoning
In turn these broke down as follows:
- There were 968 regions and of these 889 were active in at least two domains (91.8%). On average regions were active in 4.32 domains. I.e. most regions demonstrate re-use
- 555 regions were active in action tasks and 537 of these are active in 3.97 in non-action domains on average and 530 were active in cognitive domains
- There were 565 regions were active in perception tasks. 555 of these in an average of 4.00 non-perception domains; of which 550 were active in an average of 3.20 cognitive domains.
- 348 regions active in both action and perception tasks
This offers strong support for the redeployment hypothesis. And the claim becomes even stronger if one uses larger regions because the amount of re-use increases:
- e.g. of 66 regions 65 were active in two or more domains = active in an average of 9.09 different domains
- 60 regions that were active in action tasks were active in an average of 7.38 non-action domains and 5.5. cognitive domains
- 64 regions that were involved in perceptual tasks were also active in 7.39 non-perceptual domains and 5.34 cognitive domains
- 59 regions were active in both perception and action tasks were also active in an average of 5.53 other domains
Anderson takes the NICAM database as evidence to support the following claims:
- Regions of the brain are default re-used
- Regions involved in perception and action tasks are more re-used
- But other regions are still very likely to be re-used
And therefore, “…neural reuse appears to be a pervasive feature of the functional organization of the brain, and although circuits that support action and perception may be favored targets for reuse, reuse is by no means restricted to sensorimotor circuits”. I.e. neural re-use = pervasive but not limited to sensorimotor circuits, thus the need for a global theory. However, upon reviewing the four approaches to neural reuse, Anderson ultimately expresses dissatisfaction and argues that we need to explore this topic more.
The open peer-review section has a number of particularly interesting amendments and queries, which I list here briefly:
- Iriki (pp. 276-277) outlines the importance of the macaque studies and the lassitude of the body schema for neural reuse
- Both Katz (pp. 278-279); and Niven & Chittka (p. 285) argue for comparative work – especially with invertebrates – for exploring reuse in an even more refined manner. Andrew Barron at Macquarie university explores bee brains for a similar reason. He has written a really interesting paper with Colin Klein on conscious in insects (the latter incidentally points out some interesting complexities with how we use neuroimaging data (pp. 280-281)). And Barron is a really engaging speaker who is well worth checking out.
- Both Immordino-Yang and colleagues (pp. 275-276), and Michaux and colleagues (p. 278-279) argue for the importance of cultural learning – and I think that this is conspicuously missing from Anderson’s account. And at one stage, in his response, he even seems to go so far as to suggest that it would make cognitive science too difficult because it would render the unit of analysis too intractable science because “…the attendant increase in the complexity of the problem space could encourage the community to simply ignore such cases as beyond the current reach of scientific method” (p. 296).
In the related discussions:
1. On p. 261, Anderson outlines Dehaene & Cohen’s (2007) neuronal recycling position. A key component of this is the claim that reading is not very variable across cultures. Dehaene (2009) makes a number of universal claims about reading – all of these are debunked in a great paper by Max Coltheart (2014) who shows quite thoroughly that Dehaene’s view is very Western-centric, and does not take in the sheer scale of cultural variability in writing systems in the world.
2. Whilst discussing the anatomy of neural regions, the topic of split brain patients came up. Here is a great VS Ramachandran discussing a particularly intriguing case in which one hemisphere believed in god but the other was an atheist.
3. There was also an extended discussion about differences in emotions across cultural groups. Here is a quote from the previous weeks reading by Downey & Lende (2012, pp. 47-48) which might be interesting in these regards:
…a series of papers examined cortisol and other stress hormones in relation to the “ honor complex ” among Southern males ( Cohen, Nisbett, Bowdle, & Schwarz, 1996 ; D ’ Andrade, 2002 ; see also Kemper, 1990 ); the research suggests that a culturally distinctive form of masculinity involves heightened sensitivity to social slights and increased endocrine response. These findings parallel research on testosterone and challenges to status, including among nonhuman primates, which finds that the endocrine system is responsive to behavioral and emotional states (for reviews, see Archer, 2006 ; McAndrew, 2009 ; Sapolsky, 2004). Other research has explored populations with distinctive neuroendocrine and autonomic responses to basic human sensations, such as feeling compassion ( Lutz, Dunne, & Davidson, 2007 ), pain (Grant et al., 2010; Kakigi, Nakata, Inui, Hiroe et al., 2005 ), or fear ( Chiao et al., 2008a ). These cases suggest that limbic systems can be affected by patterns of early experience, and are provoked (or suppressed) by symbolic situations, social contexts, or even learned techniques for self-manipulation, such as reappraisal ( Barrett, Mesquita, Ochsner, & Gross, 2007 ; Davidson, Jackson, & Kalin, 2000 ; Ochsner & Gross, 2008; Worthman et al., 2010 ).
4. We also discussed the relation of mathematical logic and philosophy – on this topic I highly recommend the following small textbook: Eric Steinhart More Precisely: The Math You Need To Do Philosophy.
5. Lastly, we discussed a range of topics – very pertinent to me – roughly triangulated around the notion of scientific communities and how members from different schools of thought are able to communicate to one another effectively. There are many interesting issues here which I would like to briefly list:
- In relation to the excellent NICAM database Anderson and colleagues built, it is notable that there is an increasing use of online ‘Wikipedia’ style databases to try and tackle the sheer complexity of certain general and scientific problems by bringing together many agents to tackle the problem collectively (see Kittur & Kraut 2008 for an assessment). Examples pertinent to the topics of culture and cognition include Poldrack and colleagues Cognitive Atlas which is nicely explained in this open access paper. There is also Neurosynth which helps researchers conduct meta-analyses of neuro-imaging data.
- These databases point to the increasing extent to which science is becoming a collaborative or even “radically collaborative” exercise (Winsberg et al 2014). Which in turn raises questions about how knowledge is distributed (Kerr & Gelfert 2014).
- And lastly, in relation to how different scientific communities have to put in effort to communicate to one another – Peter Galison (1998, ch9) has argued that we need to create trading zones in which technical pidgins and creoles are formed as trading languages to overcome differing technical vocabularies, concerns and thought styles of different fields.
There is way more to say on each of these topics and I would be happy to discuss them if anyone would like to.
Here is a video of Mike Anderson discussing reuse and plasticity in a class context – sound isn’t great but is quite interesting as he is taking and answering questions.