Each AFSM has an input and output signal. When the input of an AFSM exceeds a predetermined threshold, the behavior of that AFSM is activated (i.e. the output is activated). The inputs of AFSMs come from sensors or other AFSMs. The outputs of an AFSM are sent to the agent's actuators or to the inputs of other AFSMs.
Each AFSM also accepts a suppression signal and an inhibition signal. A suppression signal overrides the normal input signal. An inhibition signal causes output to be completely inhibited. These signals allow behaviors to override each other so that the system can produce coherent behavior .
The use of AFSMs results in a tight coupling of perception and action, producing the highly reactive response characteristic of subsumption systems. However, all patterns of behavior in these systems are pre-wired.
AFSMs are the only units of processing in the architecture. Thus, there are no symbols.
Several additional extensions (Mataric, 1992) have been proposed to pure reactive subsumption systems. These extensions are known as behavior-based architectures. Capabilities of behavior-based systems include landmark detection and map building, learning to walk, collective behaviors with homogeneous agents, group learning with homogeneous agents, and heterogeneous agents.