The decision lobe outputs the creature's chosen action to the game engine. Each neuron corresponds to a particular action, like "approach" or "sleep".
Pre-Creatures Evolution Engine (c2e), each neuron has an input from several concept lobe neurons, whose weights represent how good or bad an idea that decision is when the situation represented by the input neuron is happening. Recently active connections have their weight lowered by punishment and raised by reward, and disappointment causes the connection to become more likely to migrate.
In the Creatures Evolution Engine games, each neuron has an immutable connection from the row of combination lobe neurons corresponding to that action.
In all games, inputs of each neuron are summed up, and the neuron with the highest activity is selected by the game engine as the action performed by the creature. Pre-c2e, this behaviour is achieved by the "winner takes all" (WTA) flag in this lobe's gene. In c2e, this is done by special SVRules.
Type 0 dendrites are links that contribute towards norns doing the specified action, type 1 dendrites are links that discourage norns from doing the specified action. source
Notes from the .gno file
The gene description file that comes with the Genetics Kit attaches the following notes to the decision lobe.
- This lobe learns relationships between concepts and appropriate/inappropriate actions. It sums the excitory & inhibitory recommendations from concept space.
- It's an interneurone lobe, because its outputs have to be combined with spoken verbs - the creature decides what he wants to do, but this may be overridden by what the speaker wants him to do.
- NOTE: Even though Decision Lobe is WTA, this one must be WTA too, because that's crucial for the assessment of susceptibility to reinforcement.
|This stub could use .|