Consciousness: The Tools
Give us the tools of 2000 generations or so of cultural evolution, joint attention, high stakes, a division of labor, and a lot of work, and we can form and reform, to a significant degree, our own consciousnesses.
Author's Note
I have written a paper that sketches out a theory of consciousness.
What I argue in this paper is that what we call—and feel as individual—human consciousness is the product of co-evolution with a strong sexual selection component. This individual consciousness is an internalization of what I call, following the ideas in (Frith, 2025) [PDF], social consciousness. The internalization of social consciousness along with social consciousness itself made it possible for Homo sapiens to migrate out of Africa—indeed, it compelled this migration possibly—some 70,000 to 50,000 years ago and marks a dividing line for the species when cultural evolution begins to operate on ourselves. Consciousness allowed our ancestors to self-domesticate and spread out while maintaining cultural alignment. It operates in the same way in human infants and adults, providing for rapid cultural and language learning and affiliation in the face of noisy environments and enormous poverties of linguistic stimuli.
This is the second installment of this series. The first is here.
II. The Tools
It is a prediction of this theory of consciousness that consciousness itself will frustrate our attempts at explaining it (a similar prediction falls out of the global workspace theory, which I linked in the previous installment). It will be 'inclined' to make our current (more or less) social and human-cultural environment seem like the default (knowledge and science must still rescue us time and again from this human-cultural-centeredness). Thus, when we look at our society today and try to imagine taking just about everything away and rebuilding human cultures and languages from scratch, we I think justifiably—but also unjustifiably—laugh. We seem to imagine that, given how 'advanced' we are (yet stupid), and how powerful our consciousnesses are presumed to be—and they are—(yet fragile and teachable), some other more-powerful-than-us force must be mostly responsible for assembling them. But I think we just need to see past this. Give us the tools of 2000 generations or so (compared with, say, 200 since the oldest writing) of cultural evolution, joint attention, high stakes, smaller numbers, a division of labor, and a lot of work, and we can form and reform, to a significant degree, our own consciousnesses.
Another tool we cannot forget is, with joint attention, inferential alignment. Again, these people were a lot like us: they didn't need 'language' to communicate in closely affiliated groups—meeting another's glance and laughing (a noise made by some mammals during rough play to indicate that "we aren't taking this seriously"), emotional displays, physical violence and coercion, winks, smiles, etc. None of these signs are strictly necessary, since the 'content' of this kind of communication is a feeling—mutual feelings.
This first-channel inferential communication 'system' is still with us in assumedly all its force, and it provides for yet another helpful non-linguistic tool for culture-building (besides itself): imitation. Just as in Dor's theory of language development and use, where fuzzy, private "analog" feelings are converted into "digital," shared, mutually negotiated, coordinative meanings, so our Paleolithic tribes had to 'suddenly' take their already-existing fuzzy, conversationally 'private,' analog inferential alignments and 'conversationalize' them—coordinate and collaborate more discretely (and more accountably) around them. This inferential channel of communication—again, still with us—also compels us to revisit and reform what has been conversationalized, profitably or not, on occasion.
Thus, when I talk about the transmission model of communication, I am referring to the Shannon-Weaver transmission model of communication with adjustments. I assume, for instance, that inferential alignments can travel around freely in this model and that speakers and listeners actively make them during and outside of conversation—it's just that implications and inferences don't normatively travel around much when you move to optimize for coordinative conversation. Another adjustment that we must make to the Shannon-Weaver model is that we have to add turn-taking, as studied in conversation analysis.
Finally, in this second section of the paper, I theorize that the action of listener selectivity shaped the kinds of mutual agreements we made, and of course still do. This will, of course, necessitate bringing sexual selection and biological gender into the discussion next time. Both are necessary for the theory as outlined, of course, but I think we should not be afraid to see that gender lies close to consciousness yet it doesn't control it.
As a final note, here is an old tune that is all about the power of listener selectivity. It's called the Old Woman's Courtship, a call-and-response folk song that you may (or likely not) remember.
And here is Section 2 of the paper:
To look at conversation today through the lens of the transmission model, one finds a markedly burdensome role for the speaker and a powerful one for the listener, cross-culturally. The speaker must generate almost the entire proposal for shared agreement herself, and she must follow the rules the listener expects for message formation (grammaticality, etc.). Yet, on the other hand, a simple agreement or rejection is all that is required from the listener (besides trying to reconstruct the transmitted experience). Without listeners' agreements, shared meanings don't get created, no matter how brilliant the speaker thinks his ideas are or how much work she puts into it. The listener is under very little normative obligation to respond to the speaker at all except with agreement, non-agreement, or repair suggestion: huh? The listener thus has an obvious power of selectivity. As an analogy to contract law—another way we have collectively coordinated our speech and behavior—it is the offeree, not the offeror, who has the power to execute an agreement (the 'power of acceptance'). The speaker is the "master of the offer," who must spell out all the necessary and important terms in order to induce agreement from the offeree. But if there's no acceptance (which can sometimes be made silently), there's no contract.
If the listener role is selective and (assumedly) has been, this selectivity makes sense in a world governed by natural selection, where we find ourselves awash in random mutations as well as 'random' talk and thoughts, new and old ideas, and unconversationalized inferential alignments. Selectivity also makes sense for any group that must live under the laws of entropy, where 'beneficial' things are difficult to come by yet easy to destroy: you should always bet on a fire getting put out rather than one getting started. In other words, talk is cheap—plentiful and unlikely to be worth the investment of listening.
Listener selectivity is thus important, but it does not yet have any character. Going back to our pre-linguistic brethren and sistren, trying to coordinate themselves under a shared communication system, we must assume that listener interpretation is as unreliable as speaker message. Speaker wrongness is always with us, but, when we are trying to create or learn a language from the beginning, listener wrongness (struggling to interpret what to me is a pidgin of grunts, gestures, and semi-expected proto-words) will at least assumedly match speaker error. How could the listener selectively improve the speaker's message when she is just as likely to be wrong in her selection as the speaker is to be wrong about the world?
The optimal way to do this over time may be to shape the speaker (and as a byproduct, of course, the listener) into greater affiliative alignment with the listener. If the listener, in some sense, has only a spongiform notion of what you're talking about, his selection of message will naturally strongly favor those messages (speakers) that bear more markers of affiliativity, alignment, or familiarity—messages with more parts that are 'familiar' to his experience, or perhaps 'affiliatively aligned with his inferences' (because affiliatively aligned inferences and feelings of familiarity are all we [as primitive humans and as children, pre-linguistically, with joint attention] had). In addition, selecting for affiliatively aligned speakers-thoughts-messages mitigates the error asymmetry between speaker and listener: a speaker's error is more likely to be spread wide than the private interpretational errors of a listener. It is, then as today, far worse for forty-five people to inferentially align around a single stupid idea than it is for forty-five people to each individually (and thus uncoordinatedly) believe forty-five separately stupid things.
