Experiment 2: linear and parabolic accelerations

(a). Stimuli

This experiment did not enq)loy an adaptive procedure for the following reasons. Firstly, the minimum number of 100 stimuli required (2 profiles x 50 stimuli) was not compatible with a reasonable test duration. Secondly, due to chair stiction on the track, the multiplication of a reference waveform would have produced inaccurate acceleration onsets which would have been sensed as small jolts. Thirdly, the limited track length of 7 meters prevented the use of suprathreshold accelerations which are necessary with adaptive psychophysical techniques, bearing in mind that smooth motion onsets had to be delivered.

Instead of using an adaptive procedure, 3 fixed waveforms, 2 linear and 1 parabolic accelerations, were generated. They all included a signal compensating for the chair stiction, thus assuring smooth and controlled motion onsets: this signal lasting 1.5 s was conq)osed of three velocity ramps which created slight chair vibrations without generating motion. The linear stimuli comprised a lower acceleration ramp (SlowR) which attained 28.4 cm/s^ and 1.4 m/s in 9.9 s, constant gradient of 2.8 cm/s^, and a higher acceleration ramp (FastR) which attained 56.2 cm/s^ and 2 m/s in 7.1 s, constant gradient of 7.9 cm/s^. The parabolic waveform (Par) produced an acceleration of 60.4 cm/s^ and a velocity of 1.83 m/s in 8.9 s, with a gradient rising from 0 to 13.5 cm/s^, thus providing the smoothest motion onset and intersecting the gradients of the 2 linear accelerations. These stimuli reached high acceleration levels as preliminary experiments on normal subjects showed that motion detection failed for stimuli with smooth onsets and final accelerations lower than 20 cm/s^. The three acceleration profiles were randomly ordered, and presented ten times to the subjects: 5 times in each direction of motion.

To reach strong accelerations while keeping gentle motion onsets, the maximum length of the track was used. Therefore, before each stimulus, the chair was driven to one end of the track without the subjects being informed of their location. This was done by using a subliminal acceleration consisting of a single cycle sinewave (bell shape for the velocity signal) with a peak value of 2.7 cm/s^, a period of 25 s, and a maximum velocity of 20 cm/s. During these ‘positioning’ stimuli, the subjects were asked to count the number of flashes of a head-fixed light emitting diode (LED) situated at 40 cm. The LED, switched on and off with a square wave whose frequency varied between 0.4 and 1.4 Hz, reminded the subjects not to indicate their motion direction, and also distracted them from trying to guess their final location. One might suggest that a visual cue could lower the thresholds of motion perception, in which case the accelerations of 2.7 cm/s^ used during the positioning’ stimuli might have been too high. However, the LED could safely be used as Benson and Brown (1989) have shown that head-fixed visual displays do not lower the detection thresholds of linear motion.

Chapter Three - Perception of linear motion

Velocity Acceleration Gradient of acceleration

5 cm's 15cm/s^

100 cm's 30cn^s^ 8 cnVs^

2 s

Fig. 3-3: Velocities (1^ column), accelerations (2'“' column) and gradients o f acceleration (3*^** column) associated to acceleration steps row, example at one acceleration level), linear accelerations (2"*^ row, SlowR and 3"* row, FastR) and parabolic acceleration (4* row. Par). The velocity signals are recordings o f the chair velocity feedback, and were used to estimate the associated accelerations and their gradients which were low-pass filtered at 1 Hz for noise removal.

To estimate the subjects’ reaction times, the following stimulus was presented four times during testing: slow onset comparable to the other stimuli but followed after approximately 2 s by a sudden acceleration step of 1.9 m/s^. The average delay between this sudden acceleration and the motion of the joystick was considered as the ‘reaction time’ of the subject. It was calculated to try to take into account the inter-subject variability on the time elapsed between feeling of motion direction and indication.

Two 5-minute breaks were included during the test to keep the subjects alert and concentrating on their task.

Chapter Three - Perception of linear motion

(b). Threshold determination

The analysis techniques differed from the one employed for acceleration steps as ejq)eriments with linear and parabolic accelerations did not employ adaptive psychophysical procedures. Thresholds were determined in two ways: Analysis 1 determined the time when the subjects had indicated 67% of correct responses whereas Analysis 2 was based on the average latencies of the subjects’ indications (including the incorrect ones). For both analyses, rightward and leftward displacements were combined to obtain 10 responses for each subject and stimulus profile.

A nalysis 1; thresholds defined as acceleration levels corresponding to 67% of correct detection

For analogy purposes with e?q>eriment 1, the acceleration thresholds were defined as the acceleration levels yielding 67% of correct detections of motion direction. Only responses obtained before chair deceleration were considered. For each acceleration profile, the subject’s ‘reaction time’ was subtracted from the times at which 6 and 7 directions of motion had been correctly recognised. The accelerations (Acc6 and AccT) corresponding to the two calculated times were obtained from the average chair velocity feedback. The threshold was defined as: Acc6 + 0.67 x

(Acc7 -Acc6).

Analysis 2: thresholds defined from response latencies

For each subject, the proportions of correct indications taking into account initial responses only and the mean latencies of the responses including those incorrect were determined. The subject's reaction time was subtracted from these mean latencies to yield estimates of the times at which motion direction was detected. From the chair velocity feedback, the accelerations associated to these estimated times were determined and defined as thresholds.