2.7Learning Communication Protocols
A communication protocol assigns behavioural meaning to messages: a sender chooses a signal and a receiver changes its action in response.
In this section you will
- Read a hand-designed message semantics
- Describe a protocol learned from the shared task reward
- Explain meaning through the effect a symbol has on the receiver
Hand-Designed Protocols
Section titled “Hand-Designed Protocols”The straightforward option: a human fixes the vocabulary.
| Symbol | Meaning, assigned by us |
|---|---|
0 | no request |
1 | get an ingredient |
2 | begin cooking |
3 | serve the dish |
Now the agents have one job left: learn when to use each symbol, and how
to respond. The semantics are given, so the receiver’s side is partly solved
before training starts. Send 1 and the receiver already knows it concerns
ingredients.
This is a perfectly respectable engineering choice. It is inspectable, it is debuggable, and where you genuinely know what needs saying it is hard to beat.
Its limit is that you have to know. A hand-designed vocabulary can only express distinctions its designer thought of, and the useful distinction in a task is often not obvious in advance. This is the same objection as hand-coding a policy: it works exactly as far as your understanding of the problem does.
Learned Protocols
Section titled “Learned Protocols”Now remove the meanings and keep the symbols.
- the discrete message space
- symbols whose meanings are learned through their effects
At the start of training, 2 means nothing at all. It is an integer the
sender may emit and the receiver may condition on, and that is the whole of
its definition.
What can happen during learning is this. The sender occasionally emits 2. On
some of those steps the receiver, also learning, happens to respond in a way
that works out. The team return goes up, both policies are reinforced, and the
pairing of this situation with this symbol with that response becomes
slightly more likely. Repeat, and a regularity forms.
The textbook is explicit about why this is hard: because agents do not initially know the meaning of communication actions, they have to learn how to interpret the messages of others. Sender and receiver are learning against each other, so neither side can be correct first.
Meaning Through Receiver Behaviour
Section titled “Meaning Through Receiver Behaviour”Suppose the regularity forms. Here is what it looks like, and what it does not mean.
| Situation | Symbol sent | Situation | Symbol sent | |
|---|---|---|---|---|
| ingredient needed | 3 | ingredient needed | 2 | |
| ingredient needed | 0 | ingredient needed | 2 | |
| ingredient needed | 2 | ingredient needed | 2 | |
| early in training | no stable mapping | later | a convention |
The left column is noise: the same situation produces different symbols, because nothing has settled. The right column is a protocol: a stable mapping from situations to symbols, shared by sender and receiver.
Now the sentence worth keeping from this section.
Two consequences, and both matter later.
The labels are arbitrary. Nothing distinguishes the protocol above from
one where 1 plays the role of 2. Any permutation of the symbols works
equally well, provided both agents permute together.
The meaning lives in the pair, not in either agent. There is no fact about
agent 1 that makes 2 mean anything. The meaning is a property of the
relationship, and it exists only while both halves agree.
That second point is where this chapter starts turning into the next one. A protocol is a convention held jointly by the agents that learned it, and a convention held jointly is exactly the kind of thing that fails when one partner is replaced.
Knowledge check
Correct.
Not quite.
It is lost. The symbol meant something only because of how the receiver responded, and the new receiver has no reason to respond the same way.
Correct, and this is the whole point of "meaning through use". Nothing inside the sender encodes the semantics. The frozen sender will keep emitting 2 in the same situations, and until the new receiver learns what to do about it, the symbol carries no information anyone can act on.
It is preserved, since the sender still emits 2 in the same situations.
The sender’s behaviour is preserved; the meaning is not. A symbol reliably emitted and not understood is not communication, it is noise with good timing.
It is preserved, because 2 is a fixed part of the message space.
The message space fixes which symbols exist, not what they refer to. That is the difference between a hand-designed vocabulary and a learned one.
The new receiver will infer the meaning from the sender’s behaviour.
It can, eventually, and only by learning: it has to try responses and find which ones the team return rewards. That is retraining, not inference at deployment, and the Adapt chapter is about how much of it can be done on the fly.
Explanation
Holding “meaning lives in the pair” makes the transition to the Adapt chapter inevitable rather than surprising.
Learned Communication Summary
Section titled “Learned Communication Summary”- Hand-designed messages fix the semantics, so agents only learn when to use each symbol. Inspectable and debuggable, and limited to distinctions the designer anticipated.
- Learned messages start meaningless. The symbol space is given; the meanings are not.
- Learning them is a bootstrapping problem: a message is useful only if the receiver responds usefully, and the receiver learns that only from useful messages.
- A protocol is a stable mapping from situations to symbols, shared by sender and receiver.
- A symbol means whatever its listeners do about it. The labels are arbitrary, and the meaning is a property of the pair rather than of either agent.