K expresses the structural capacity required for behaviour. It is the axis used within CCM to classify what must persist internally for a given behavioural class to be possible.
Rather than measuring intelligence or complexity, K specifies the internal processing and persistence commitments required for behaviour, from rigid same-input responses through to learning, regime switching, and explicit internal modelling. In practical terms: what must exist internally for this behaviour to be possible at all?
K levels are read as bounded architectural commitments, describing what must exist internally for the behaviour to occur, not how impressive an organism appears.
K is a capability axis, not an intelligence ladder. A higher K level means that more demanding internal structure must be present for the behaviour under consideration.
Read K by asking what new behavioural flexibility appears at each step: from identical responses, to sensitivity shifts, to differentiated responses, then internal-state control, outcome-based learning, regime switching, and finally explicit evaluation of future outcomes.
The K scale matters because distinct forms of behavioural organisation are often grouped together under broad terms such as “intelligence”, “complexity”, or “learning”. But identical response, recent-history modulation, differentiated treatment of inputs, internal-state control, outcome-based updating, regime change, switching, and explicit internal modelling are not the same thing.
By separating these levels, K helps make comparison more disciplined. It clarifies what class of internal capability is actually being implied when we attribute a behaviour to an organism or system.
This visual summarises the current public interpretation of the K scale, from same input, same response through recent-history effects, differentiated responses to different inputs, internal-state control, outcome-based updating, regime-level shifts, switching between competing regimes, and finally explicit internal modelling.
The following definitions specify the K levels as bounded structural commitments, indicating what must exist internally for each class of behaviour.
K is about what must persist internally across time for behaviour to be shaped by more than the immediate present. At lower levels this may amount to little or no retained variation. At higher levels it includes state that can bias behaviour, differentiate between input classes, encode prior outcomes, support regime switching, or contribute to explicit internal modelling.
The important point is functional rather than anatomical: not merely that some substrate exists, but that it supports behaviourally relevant carry-forward from one moment to the next.
The K scale does not assume that organisms possess rich world models at low levels. Most of the scale is about simpler forms of persistence and behavioural organisation: repeated-response effects, input-specific differences, internal-state bias, learning from outcomes, and mode switching.
Only at the highest end of the illustrated public interpretation does K move into behaviour that is best explained by explicit internal modelling of possible future outcomes.