An alternative hypothesis for the evolutionary and neurogenerative development of recursion (or inclusive hierarchies) in homo sapiens may be due to the unique social development of the species and the selective pressures toward the internal mapping of complex social structures. While complex spatial navigation is not unique to homo sapiens (avian species particularly traverse great and complex distances, as well as many migratory land and water-based animals), the language and reasoning we attribute to intelligence (an to which the concept of recursion plays a critical role) certainly is. Similarly, the complex social structures that emerged as pre-homo sapiens group size increased has been attributed to the evolutionary increase in frontal cortex size. I haven't cited any support for this substantial claim, but I do recall this from some of my undergraduate psychology research (could probably dig up some sources).
There is some literature, although not much and not particularly current, about recursion as a function of social structure mapping. Something I may dig into more:
One thing to keep in mind: it is clear that complex language, society and tools all benefit from recursion. It is also clear that all three have advanced significantly in the past 100,000 years. But that is no argument for treating any one of the three as the _origin_ of recursion.
The nature of evolution is such that any change to an organism must be beneficial to the survival of the organism long enough for the corresponding genes to take hold in the population. Furthermore, such changes must be incremental in a genetic sense, affecting just one or two genes at a time. To have an evolutionary origin story for recursion, we have to be able to point to incremental changes that are almost immediately beneficial. Certainly language, society and tools can all satisfy the requirement of being beneficial, but we must explain how these developed incrementally (as they likely did) step-by-step. Otherwise we are just waving our hands over the essence of the problem.
There are plenty of migratory birds and animals. But migration can be managed in a variety of ways, for example, by following magnetic fields, or temperature gradients, and so on. I am not suggesting that long-range navigation necessarily forces the development of recursion, I am suggesting that the particular way that early humans may have managed long-range migration benefitted from the development of navigational facilities that happened to be adapted in such a way that they enabled recursion.
I am actually suggesting the mechanisms: first the development of an ability to recreate places from interacting categories, and second the possibility of cycles emerging from the interactions. So it isn't about long-range navigation per se generating recursion, but rather about recursion being a side effect of a particular category-based approach to place recognition and navigation.
Figure 1 of the first paper you posted (https://royalsocietypublishing.org/doi/10.1098/rstb.2012.0097) is a good image of what I am proposing: The red circle represents hierarchical embedding, and the blue represents recursion. If first you get hierarchical categories, which could arise naturally enough from the value of compressing place descriptions, then second you could have cycles emerge as an unexpected and serendipitous consequence.
One objection to my argument in favor of tool use as an origin for recursion would be to point to the same process in tools: categories are useful for compressing tool descriptions, and then you can get cycles from tools rather than places.
I'm not sure it argues much against what I am saying, though, because I am suggesting that people treat tools as though they were places. I suppose the direction of the argument is reversed from the one I offered, though.
Certainly the hypothesis I advanced is tentative and there are other possible interpretations. In addition to long-range navigation, tool use and complex social structure are two other major phenomena that occur at the same point. It's worth commenting on the positives and negatives of each of these other two phenomena as an origin story for recursion, which I'll do in individual comments as I find time to do so. I could also probably reassess the proposal I put here originally, since it's been a few months and I have had more time to reflect.
On the topic of tool use, I looked for but could not find references on how tool creation and analysis is handled in the brain. There is evidence that humans use the visual and associational areas of the temporal lobe in combination with the posterior parietal lobe (where spatial relations are represented) to _use_ tools to complete actions (Orban & Caruana, 2014: https://www.frontiersin.org/articles/10.3389/fpsyg.2014.00310/full), but that is different from an explanation of how we _understand_ the tools sufficiently to create new ones. I conjecture that we use conceptual maps in the hippocampus and beyond for this purpose, but it is a conjecture only.
On the topic of social complexity, it is worth noting that people organize their social relationships much like a navigational map (Tavares et al, 2015 -- https://www.cell.com/neuron/fulltext/S0896-6273(15)00524-3). They showed that navigating social relationships engaged the hippocampus.
I do not know whether other mammals use the hippocampus this way, but the evolutionary origin of the hippocampus is for place navigation. So the management of social complexity may arise out of the navigational system -- which is at least some evidence in favor of my point of view.
It's interesting to consider recursion as an emergent survival and fecundity mechanism and the idea a single mental and neurobiological mechanism, even further that this mechanism involves place maps.
Right along with figuring out a neuroscience experiment to determine if places can be treated as objects in the brain, it would be interesting to come up with non-human experiments on recursion in objects and places to trace whether there is something like a evolutionary path in non-verbal thinking. I'd imagine there is, but how to measure it seems very tricky.
I enjoyed the visualization on the point of saccades - the idea of "thinking" as a kind of conceptual saccade. I always visualized embedded search (e.g. FAISS) as bouncing around multidimensional space like this.
Thinking about LeCun's recent blog post (https://ai.facebook.com/blog/yann-lecun-advances-in-ai-research/) in context to the notion of binding as a general property of feature propagation through concept hierarchies during navigation, a control/configurator module with a world module as necessary components to intelligence seems to agree with this proposition. This gets right to your point about a control policy. Taking the analogy a bit further, maybe something like FAISS is one of the more basic algorithms representing control and world modules for conceptual traversal.
Thinking about your discussion on syntax as analogy of traversal, transformer models come to mind. What could be said about modern transformers' similarity to place maps in this context?
An alternative hypothesis for the evolutionary and neurogenerative development of recursion (or inclusive hierarchies) in homo sapiens may be due to the unique social development of the species and the selective pressures toward the internal mapping of complex social structures. While complex spatial navigation is not unique to homo sapiens (avian species particularly traverse great and complex distances, as well as many migratory land and water-based animals), the language and reasoning we attribute to intelligence (an to which the concept of recursion plays a critical role) certainly is. Similarly, the complex social structures that emerged as pre-homo sapiens group size increased has been attributed to the evolutionary increase in frontal cortex size. I haven't cited any support for this substantial claim, but I do recall this from some of my undergraduate psychology research (could probably dig up some sources).
There is some literature, although not much and not particularly current, about recursion as a function of social structure mapping. Something I may dig into more:
- Distinctive signatures of recursion (https://royalsocietypublishing.org/doi/10.1098/rstb.2012.0097)
- Challenges for complexity measures: A perspective from social dynamics and collective social computation (https://aip.scitation.org/doi/full/10.1063/1.3643063)
- Reflexivity, recursion and social life: elements for a postmodern sociology (https://journals.sagepub.com/doi/10.1111/j.1467-954X.1989.tb00048.x)
One thing to keep in mind: it is clear that complex language, society and tools all benefit from recursion. It is also clear that all three have advanced significantly in the past 100,000 years. But that is no argument for treating any one of the three as the _origin_ of recursion.
The nature of evolution is such that any change to an organism must be beneficial to the survival of the organism long enough for the corresponding genes to take hold in the population. Furthermore, such changes must be incremental in a genetic sense, affecting just one or two genes at a time. To have an evolutionary origin story for recursion, we have to be able to point to incremental changes that are almost immediately beneficial. Certainly language, society and tools can all satisfy the requirement of being beneficial, but we must explain how these developed incrementally (as they likely did) step-by-step. Otherwise we are just waving our hands over the essence of the problem.
There are plenty of migratory birds and animals. But migration can be managed in a variety of ways, for example, by following magnetic fields, or temperature gradients, and so on. I am not suggesting that long-range navigation necessarily forces the development of recursion, I am suggesting that the particular way that early humans may have managed long-range migration benefitted from the development of navigational facilities that happened to be adapted in such a way that they enabled recursion.
I am actually suggesting the mechanisms: first the development of an ability to recreate places from interacting categories, and second the possibility of cycles emerging from the interactions. So it isn't about long-range navigation per se generating recursion, but rather about recursion being a side effect of a particular category-based approach to place recognition and navigation.
Figure 1 of the first paper you posted (https://royalsocietypublishing.org/doi/10.1098/rstb.2012.0097) is a good image of what I am proposing: The red circle represents hierarchical embedding, and the blue represents recursion. If first you get hierarchical categories, which could arise naturally enough from the value of compressing place descriptions, then second you could have cycles emerge as an unexpected and serendipitous consequence.
One objection to my argument in favor of tool use as an origin for recursion would be to point to the same process in tools: categories are useful for compressing tool descriptions, and then you can get cycles from tools rather than places.
I'm not sure it argues much against what I am saying, though, because I am suggesting that people treat tools as though they were places. I suppose the direction of the argument is reversed from the one I offered, though.
Certainly the hypothesis I advanced is tentative and there are other possible interpretations. In addition to long-range navigation, tool use and complex social structure are two other major phenomena that occur at the same point. It's worth commenting on the positives and negatives of each of these other two phenomena as an origin story for recursion, which I'll do in individual comments as I find time to do so. I could also probably reassess the proposal I put here originally, since it's been a few months and I have had more time to reflect.
On the topic of tool use, I looked for but could not find references on how tool creation and analysis is handled in the brain. There is evidence that humans use the visual and associational areas of the temporal lobe in combination with the posterior parietal lobe (where spatial relations are represented) to _use_ tools to complete actions (Orban & Caruana, 2014: https://www.frontiersin.org/articles/10.3389/fpsyg.2014.00310/full), but that is different from an explanation of how we _understand_ the tools sufficiently to create new ones. I conjecture that we use conceptual maps in the hippocampus and beyond for this purpose, but it is a conjecture only.
On the topic of social complexity, it is worth noting that people organize their social relationships much like a navigational map (Tavares et al, 2015 -- https://www.cell.com/neuron/fulltext/S0896-6273(15)00524-3). They showed that navigating social relationships engaged the hippocampus.
I do not know whether other mammals use the hippocampus this way, but the evolutionary origin of the hippocampus is for place navigation. So the management of social complexity may arise out of the navigational system -- which is at least some evidence in favor of my point of view.
It's interesting to consider recursion as an emergent survival and fecundity mechanism and the idea a single mental and neurobiological mechanism, even further that this mechanism involves place maps.
Right along with figuring out a neuroscience experiment to determine if places can be treated as objects in the brain, it would be interesting to come up with non-human experiments on recursion in objects and places to trace whether there is something like a evolutionary path in non-verbal thinking. I'd imagine there is, but how to measure it seems very tricky.
I enjoyed the visualization on the point of saccades - the idea of "thinking" as a kind of conceptual saccade. I always visualized embedded search (e.g. FAISS) as bouncing around multidimensional space like this.
Thinking about LeCun's recent blog post (https://ai.facebook.com/blog/yann-lecun-advances-in-ai-research/) in context to the notion of binding as a general property of feature propagation through concept hierarchies during navigation, a control/configurator module with a world module as necessary components to intelligence seems to agree with this proposition. This gets right to your point about a control policy. Taking the analogy a bit further, maybe something like FAISS is one of the more basic algorithms representing control and world modules for conceptual traversal.
Thinking about your discussion on syntax as analogy of traversal, transformer models come to mind. What could be said about modern transformers' similarity to place maps in this context?