Estimates of time to collision based on binocular and monocular visual information

Purpose: To measure the absolute accuracy with which observers can estimate time to collision using binocular information alone (BI), monocular information alone (MI) and combined binocular and monocular information (MI+BI). Methods: An approaching object was simulated by changing the disparity and/or size of a spot presented on CRT monitors A brief tone was presented a fixed interval after stimulus onset. The observer's task was to judge whether the object would have collided with his/her head before or after the tone. The time to collision of the stimulus was varied according to a staircase method. Four starting sizes and four starting disparities were randomly interleaved. Results: Using BI, errors in estimates were small (<6%). Errors in estimates based on MI were small (<3%) for the two smallest starting sizes (0.4 and 0.6 deg) but were larger (>6%) for the largest starting size (1.2 deg) Errors in estimates based on MI+BI were significantly lower than for MI or BI alone and observers showed no effect of starting size Conclusions: 1) BI can be used to judge absolute time to collision accurately, at least for the close viewing distance (1.6 m) we used. 2) MI can be used to judge time to collision accurately, though observers show a small effect of starting size. 3) Judgements are significantly more accurate and there is no effect of starting size when MI+BI is used.