Reflexes take advantage of the most low-level features of the BrainStem.  Using them successfully requires thorough knowledge of the command set and device I/O.

The reflex architecture is a mechanism by which one command or input can trigger an additional command or series of commands.  This is done with reflex vectors and reflex commands stored in the EEPROM.

Reflexes allow you to configure and manipulate automatic responses to various input changes within the BrainStem network.  An example might be stopping the motors of a mobile robot when a bumper switch is depressed.

Reflexes can be configured to happen automatically with no intervention from the host.  When the host is unreachable or busy with a high-level task, a reflex can monitor inputs and take appropriate actions when necessary.  When the host is ready, it can check the BrainStem network to determine its state.

Since not all reflexes are applicable to all the states a robot may encounter, it is easy for the host to modify, disable, or redirect reflexive actions as the robot's states are changing.