19/09/2026

A Unified Theory of Protomusic, Music and Speech as Systems of Communicative Thought Control

In June 2023 I published https://philipdorrell.substack.com/p/speech-as-a-form-of-thought-control (also now republished in this blog at https://blog.thinkinghard.com/blog/speech-as-a-form-of-though-control).

The main idea is that a speaker controls the thoughts of the listener, and that this is an essential and unavoidable part of what it is to be a listener listening and understanding what someone is saying to you.

We might wonder why this should be the case. Should it not be possible to receive information without at the same time being controlled in some way by the person sending the information?

Surely, if evolution could find a way to separate the two things, it would have done so.

So we could conclude that, for some reason, there is a strong constraint preventing that separation.

That is, it’s almost an axiom:

Comparing to Non-Biological Communication

In the modern world of electronic computers, it seems fairly obvious that Computer X can receive communications from Computer Y, and store the content of those communications, and process them in various ways, without Computer X being required to “believe” in any sense that the content of any communication is true.

Also, these days, there is no significant cost for a computer to receive and store communications corresponding to the contents of an ordinary human conversation.

Actually, there is a very modern exception to these observations, which has to do with prompt injection.

If I feed something like this to an LLM chat UI:

It might actually give me the second sentence in Spanish, not Portuguese.

And if it doesn’t, it is because the underlying model has been trained to detect prompt injection attempts and not to fall for them.

Now LLMs are created by, in some sense, copying the observed behaviour of thousands of millions of people writing stuff on the Internet and elsewhere, so we could argue that any difficulty they have in receiving information without at the same time determining that the information is true might be the result of being copied from the behaviour of actual biological organisms.

Nevertheless, despite the specific LLM example, the overall conclusion is, that if we try to extend the above “axiom” to include all communication systems, not just biological systems, then it is obviously not true.

Mitigations

Even if there is no way to prevent a listener from having their thoughts “controlled” in the sense of believing the truth of whatever is being said by a speaker, we would still expect the evolution of mitigations, in particular to reduce the risk of believing the truth of something that is possibly or probably false, whether that be due to the intention of the speaker, or just due to the speaker being wrong for whatever reason.

The simplest and most direct form of mitigation is to evaluate everything a speaker says as soon as it has been said. So in the worst cases, the speaker is “controlling” the listener’s thoughts, but if the speaker successfully and critically determines that what the speaker says is false, or at least implausible and unverifiable, then said control may be very short-lived and transient.

Of course the process of evaluation is not cost-free, and may involve significant mental effort in some cases.

So the only way to completely avoid paying a “cost” due to listening to a speaker saying things that are not true is to actually never be close enough to that speaker to hear what they say.

Comprehension vs Planning to Act

The axiom above can be extended to include a secondary axiom (or perhaps we could consider it a corollary).

If we learn a new fact, and believe it to be true, then our belief will often require us to formulate some action with regard to that new fact. In some cases the requirement to act will be quite urgent.

So the revised axiom is:

This adds to the cost of receiving information, in the case where the information is false or dubious. While the truth of the received information is being evaluated, not only is the listener temporarily believing that information to be true, but, due to that belief, they are starting to think about how they should respond to that new information.

Protomusic and Music

So far I have talked about how these axioms apply to the comprehension and response to speech.

But the axiom as stated applies to all systems of biological communication between different individuals.

As part of my attempts to explain what music is and how it evolved, I have developed the hypothesis that music seems like a form of communication, but actually it is not a pragmatic system of communication, but, originally it evolved from an ancestor that was a pragmatic communication system.

Also, in its very earliest form, protomusic was a system for communicating emotions.

So applying the axiom to the hypothesised protomusic, we must conclude that protomusic was a system of thought control where the sender controlled the emotions of the receiver. In other words, if the sender made a protomusical vocalisation with a sad meaning, this directly caused the receiver to feel sad.

In effect, the sender was saying “I feel sad about something that is happening or perhaps something that has just happened”, and you the listener (or listeners) should also feel sad about that.

Evolution by Mitigation

Next we can consider what it means for communicative proto-music to have evolved into non-communicative music.

We could just suppose that protomusic evolved to be less thought-controlling.

However, an obvious effect of music in its modern form is that it does have some effect on the mental state of the listener, although we would probably stop short of saying that it controls the listeners’ thoughts.

An alternative to assuming a simple reduction in the degree of thought control is to apply the concept of mitigation as described here, and suppose that communicative protomusic evolved into non-communicative music by ramping up the level of mitigation, in effect by responding to all musical “communications” by assuming as quickly as possible that they are very probably not true.

This idea, as it happens, fits in with some of the specific observations that I have previously made about the nature of our response to music.

In particular:

Conclusion

The final conclusion of this analysis is that the speech, protomusic and music can all be understood in terms of what kind of thought control is or was involved in those systems: