Olfactory Discrimination

The automated olfactory discrimination task provides a rigorous, operant-based assessment of olfactory sensory function, perceptual acuity, and odor learning in rodents. ConductMaze controls a computer-driven olfactometer that delivers calibrated odorant stimuli through an array of solenoid-gated vaporization channels converging on a central nose-poke port. On each trial, the animal initiates stimulus delivery by inserting its nose into the odor port (detected by infrared beam break), whereupon ConductMaze opens the appropriate solenoid valve to present either the S+ (rewarded) odorant or the S- (unrewarded) odorant in pseudorandom sequence. For S+ trials, the animal must proceed to the reward port to collect a water or sucrose reward within a limited response window; for S- trials, the correct response is to withhold and return to the odor port. This go/no-go design isolates olfactory discrimination from motor or motivational confounds.

The olfactometer achieves precise concentration control through mass flow controllers that dilute saturated odorant vapor with clean carrier air to target concentrations (typically 0.01-1.0% saturated vapor). ConductMaze manages the full stimulus matrix: odorant identity, concentration, delivery duration (typically 1-2 seconds), and inter-trial interval. The system supports multi-odorant batteries for olfactory screening panels, concentration-response series for threshold determination, and reversal schedules where S+ and S- assignments are swapped to test cognitive flexibility. This paradigm is essential for evaluating olfactory deficits in Alzheimer disease models (APP/PS1, 3xTg-AD), olfactory bulbectomy as a depression model, post-viral anosmia studies, and Parkinson disease models where olfactory loss precedes motor symptoms.

ConductMaze logs every trial with odorant identity, concentration, response (hit, miss, correct rejection, false alarm), response latency, odor sampling time (nose-poke duration during stimulus), and reward collection latency. The software computes signal detection metrics including d-prime (discriminability), response bias (criterion c), discrimination ratio (hits / [hits + false alarms]), and learning curves fitted with logistic functions to extract trials-to-criterion. Reversal learning performance is analyzed separately, yielding perseverative error counts and reversal-specific d-prime. Automated session progression advances animals through shaping, acquisition, and reversal phases without experimenter intervention.