Abstract
Some blind humans actively echolocate, analyzing reflections from self-generated vocalizations to perceive the environment. Similar to visual scene exploration, real-world echolocation is an active-sensing process, requiring the goal-directed interplay of motor actions (e.g. locomotion and oral click production) and the resulting sensory percepts. To explore this dynamic, here we recorded head movements and clicking behavior of participants as they echoacoustically localized a target object. In each trial, participants used echolocation to guide their head orientation toward the target, positioned at a 1-meter distance and random azimuth in the frontal hemifield. Target size was Large or Small to manipulate difficulty; control blocks omitted active clicks. Participants ended the trial by holding their head still and pressing a button. Analysis of video recordings extracted frame-by-frame information such as target azimuth, head pose, azimuth error (head/target relative angle), and the participant's response (head orientation during button press). To evaluate performance and sensorimotor dynamics for each trial, we examined the final azimuth error, head orientation time course, and click timestamps extracted from the audio recording. Initial results from an early-blind proficient echolocator show clear advantages to echolocating a larger target, with average angular error of ~9.5°, vs. ~24.5° for the small target (35.8° in the passive no-click control condition). Compared to large targets, small-target trials lasted ~14s longer and contained ~11.5 more head direction reversals, both factors of 2.5, suggesting that head kinematics remain relatively invariant to task conditions. Small-target trials contained ~3.3x as many clicks on average, reflecting a moderate but significant increase in average click rate (0.44 Hz, p<.001). Pooled across trials, small-target clicking behavior converged steadily toward the target center, with over 75% of clicks directed within 20° of the target, suggesting an intensity-maximization strategy. Together, these results illustrate naturalistic sensorimotor dependencies underlying auditory active sensing in the absence of vision.