1. Acceleration is detected by comparing initial and final velocities
Brian Timney, Samantha Kearney, and Benjamin Asa.
Western University, London, Ontario, Canada

2. Visual Acceleration Necessary complement to motion perception
Virtually neglected in the literature, most of what is available is quite old
These older studies were technologically limited
We decided to do some parametric studies and attempt to gain an understanding of the mechanisms underlying acceleration perception

3. How is acceleration detected?
Acceleration may be detected in one of two ways
Direct
Thresholds determined by acceleration rate.
This implies the existence of “acceleration detectors” that process rate of change of motion directly
Indirect
Thresholds determined by ability to detect that speed has changed
This implies that the critical variable will be the difference between initial and final velocity, not the absolute acceleration rate

4. Experiment 1:Threshold acceleration rate
Speed
Time
200 ms constant velocity at beginning and end of presentation
Acceleration rate varied from trial to trial

5. Experimental 1:Threshold acceleration rate
We measured ability to detect acceleration for a range of starting velocities
General Methodology
100 black dots in 5° x 5° white square aperture
2-interval, 2 AFC; Constant velocity vs Acceleration
Constant stimuli: 6 acceleration rates, 25 trials/rate, 4 runs
Report which interval contained dots that were accelerating
Starting velocities: 1 to 32° s-2
2 presentation durations: 500ms and 1000ms

6. Acceleration Detection Performance
Performance improved with acceleration rate
Greater increases in acceleration were needed for faster starting velocities (Weber’s Law)
Acceleration thresholds varied with presentation duration

7. Performance vs Final Velocity
Original data replotted to show performance as a function of final velocity
Each pair of curves represents one starting velocity
Effect of presentation duration disappears
Threshold is dependent on the speed difference between initial and final velocity

8. Experiment 2a: Varying the acceleration rate
Experiment 1 provides good evidence for the indirect hypothesis, but a stronger test would be to decouple absolute acceleration rate from average acceleration over the acceleration period;
To do this we introduced a speed plateau during the acceleration phase;
If thresholds rely on detecting speed differences, then average acceleration over the presentation period should be constant at threshold

9. Experimental Paradigm
Representation of two different stimulus presentations with different absolute acceleration rates but the same average acceleration rate
In different sessions the duration of the plateau was varied

10. Absolute acceleration thresholds
Acceleration thresholds, plotted as a function of the absolute acceleration rates, for starting velocities of: A) 2° s-1 and B) 10° s-1
Absolute acceleration threshold increases as a function of plateau duration

11. Average acceleration thresholds
Acceleration thresholds, plotted as a function of the avaerage acceleration rates, for starting velocities of: A) 2° s-1 and B) 10° s-1
Average acceleration threshold is independent of plateau duration

12. Experiment 2b: Multiple acceleration rates
Data from Experiment 2a suggest that recognizing that a target has accelerated is dependent on the average acceleration within the presentation period
As a final test, we varied the acceleration rates during presentation even further

13. Experimental paradigm
The acceleration period was broken down into three components:
250 ms initial acceleration
500 ms middle acceleration
250 ms final acceleration (identical to initial)
The middle acceleration period could be positive (red), negative (blue), or zero (green)
Velocity
Time
Acceleration Period

14. Absolute vs. average acceleration
Thresholds vary markedly when plotted as a function of the first last acceleration phases
When data are expressed in terms of average acceleration over all three phases, thresholds are invariant

15. Conclusions
Thresholds for the detection of acceleration vary substantially depending on the temporal parameters of the stimulus presentation
When data are expressed in terms of the average velocity over the course of the presentation, thresholds remain relatively constant
The data suggest that the critical variable in the detection of acceleration is the difference between the initial and final velocities, regardless of the duration of the acceleration period
All of the results provide strong support for the “indirect” model of acceleration perception.

16. Thank You