On the Breakdown of the Bicameral Mind and the Importance of Diversity

Saturday, August 12, 2017

Have you heard of the bicameral mind? If you haven't, it's worth checking out. While I don't think it correct, I think it takes only a slight modification to get us something highly probable. In this view, like in the Bicameral setting, we don't all experience consciousness the same way. The core mediator is the differences in sophistication of theory of mind. The brain is plastic and different percepts affect both attentional priors and how language is used for self-modelling. The more complex stories we told, the better the listeners had to be at modeling those scenarios and the better the brain had to be at representing states at a meta level.

The break-down occurred through diffusion and cultural interchange leading to richer stories which in turn called for more complex modelling. The better you are at modelling others' mental states, the better you will be at modelling your own.

Note: The next paragraph is conditional on the occurrence of a breakdown like event but the core message is independent of it. The compositionality of culture and knowledge means the gains to interaction are multiplicative which in turn affect the creativity of the citizens by the widening of available mental tools and percepts. The similarity between my exposition and the bicameral mind is that like in the bicameral mind, there was a phase change resulting in an enriching of the manner in which consciousness was instantiated in humans. The difference is that this was not due to evolution or an inferiority of previous brains but rather, a side effect of learning to learn and sharing between sufficiently diverse cultures combined with/multiplied by language and yielding a more powerful theory of mind ability.

The breakdown of the bicameral mind was not one of evolution but due to the brain's plasticity. Increasing intellectual demands from the compositionality of culture, combined with a general learning engine of the brain, resulted in more flexible approaches to thought. We know language is linked to consciousness and higher order reasoning (see [1] on how metaphor shapes reasoning or [4] for how language augments numeracy as examples) and also that it's one area where the brain must learn how to learn. While the individuals were not less conscious nor less intelligent, the lack of richness of available percepts severely circumscribed their ability for ideation. I doubt there's anything falsifiable that could be said of an ancient left/right brain breakdown, but it's certainly true that available default abstractions would strongly impact a person's manner of cognizing about the world, theorizing of events, arranging causation and most importantly, reasoning about themselves. This could plausibly affect how consciousness was realized in the individual (emphasizing that there is no objective hierarchy or ordering). Although clearly a good thing, this sort of multi-cultural richness is not without its own set of hurdles.

In what way could Ideological Diversity be an Issue?

Well, differences in ability to predict mental states yields difficulty in collaboration. If two group's mental organization is so different as to make communication very exhaustive then coordination and attending to the same signals, required for positive welfare outcomes will be difficult to achieve.

The answer however, is not less diversity, it's more cooperation. Genetic, ecosystem and hypothesis diversity (Principle of Epicurus) are unequivocally good. Neural nets and even humans [2] suffer from mode collapse, a form of lack of diversity leading to reasoning errors, which can lead to suboptimal exploration and poor decisions. If the human mind is computable then a group of humans is not other than a highly bottlenecked parallel computer.

Again, considering [2], humans which share too much in common will be akin to starting a sampler near similar locations. If we seek to converge on the part of the landscape with peak probability, it pays to sample in parallel (via multiple people) from many start locations. If the brain is a sampling based explorer, with active inference to mitigate autocorrelation, then it's clear that lived experiences leads to richer exploration than any other source of diversity (and as long as it's computable, there's very good reason [5] to believe that it is a sampler in the large and possibly variational locally). Ideological diversity must be retained however, since it does provide different attentional weightings—it's just that the gains are not as large as from a diversity of parallel active samplers.

The only sense in which diversity can be considered harmful is when they make coordination more difficult (while I do not agree with their approach, I think China understands this on some level, that and economic protectionism explains many of their policies). Ideological diversity is probably especially prone to this. Ideological diversity, I should point out, is not just what politics or economic policies you adhere to but also, what kind of music and hobbies you like! Focusing on other forms of diversity can only increase ideological diversity.

However, short range correlations (siloing into isolated cultures) does not lead to interesting structures. Recall that one plausible way for the phase change in theory of mind to have occurred is from the richness in stories from cultural interaction and sharing. Resulting then, in a gain in sophistication on conscious reasoning about agents. Wider experiences leads to richer representations—though each human possesses only a degraded average over encounters—it is yet a comparatively richer set of examplar states to draw samples from. In essence, a certain group gets the "problem" of diversity exactly backwards. Diversity issues stem not from physical differences but from experiential and ideological heterogeneity, leading to different priorities and attentional weightings. Furthermore, experience is a stronger mediator of ideology than whatever labels (e.g. centrist) is currently fashionable to apply to one's self. All the kinds of diversity which concerns this certain crowd are in fact good but the kind which they view as important (as do I), provides the most trouble.

So what then? I think the answer is to raise human children emphasizing shared commonalities, more sharing of culture (yes, I'm pro cultural appropriation) and how to empathize (that is, not dehumanizing the other) with other beings. Additionally, a lot of problems are caused by fortifying identity by dredging for differences. If we can create a society such that there are sufficient avenues and means for collaboration and sharing—whether on art, stories or music, any kind of creation—the need to find meaning by belonging to something, should lessen by a large amount.

Meaning from sharp boundaries group identity reduces a human's ability to reason, replacing it with pattern matched stock responses and confabulations. This is an example of when correlations can go wrong, trapping members in equilibria even if the members could have long since drifted past the original goals and utilities.

Prediction: If children all over the world grow up watching each other's culture's cartoons, this will have a significant ameliorative effect on any ill coordinative effects of cultural diversity. Of course, some might question the extent to which this is a prediction given the broad arch of history.

It is unlikely that humanity cannot learn to cooperate broadly and arbitrarily; this is rather, a problem of learning how to escape pernicious attractor states and achieving positive social utility through mechanism design.


In this essay, I point out that a computable brain performing sampling based inference as suggested in [2], and a group of humans yields a parallel sampler. Connecting this with the observation in [3] that autocorrelation is minimized by active inference tells us that to get better exploration and representative posterior probability distributions with respect to reality, we want a wider range of physical experiences. As a result, diversity is important. I point out however, that since ideological diversity is cognitive and attentional, it is broader than politics—increasing general diversity must increase ideology. Secondly, it can be negative, due to a reduced shared ability to attend to the same signals and efficiently communicate (coming off such different attentional weights).

I argue however, that diversity is worth it by pointing out how a scenario like the break-down of the bicameral mind might have been precipitated by trade and cultural diffusion leading to richer stories and hence better and more practicing of theory of mind (I suggested replacing origin of consciousness with enriching of consciousness, based on the compositionality of culture together with plasticity from our ability to learn how to learn). I also suggest cultural appropriation and lesser emphasis on identity/belonging and more on collaboration to counter ideological clashes of diversity.

I bring up the issue of ideologies because careful manipulation of attended to signals can lead to feedback loops which sort matched humans into clusters of sharper and wider boundaries, divorced from reality—a general human failing which gets in the way of coordination—and therefore handicaping our ability to achieve positive welfare outcomes globally. The issue is tribal (geographically correlated attentional priors and utilities) in nature and happens in every country with multiple ethnicities. Which is almost every country. I'll note that the EU was a response to one such.


[1] http://www.sciencedirect.com/science/article/pii/S1364661317301535

[2] http://www.sciencedirect.com/science/article/pii/S1364661316301565

[3] http://www.biorxiv.org/content/biorxiv/early/2017/02/17/109355.full.pdf

[4] http://langcog.stanford.edu/papers/FEFG-cognition.pdf

[5] http://www.sciencedirect.com/science/article/pii/S0278262615300038