Demonstrations of Auditory Illusions

Supervisors:
Yoshitaka Nakajima*, Takayuki Sasaki**, & Gert ten Hoopen***

Technical Staff:
Kyoko Kanafuka*, Noortje Jansen***,
Yasunori Murakita*, & Daigoh Suetomi*

Scientific Adviser:
Ger Remijn***

Editorial Assistant:
Yukiko Minenaga*

* Kyushu Institute of Design, Fukuoka, Japan
** Miyagi Gakuin Women's College, Sendai, Japan
*** Leiden University, Leiden, The Netherlands

Contents

Introduction

I. The Gap Transfer Illusion

<1> A Long Glide Tone and a Short Glide Tone Crossing Each Other
<2> A Long Continuous Glide Tone and a Short Disrupted Glide Tone Crossing Each Other
<3> The Gap Transfer Illusion
<4> A Long Glide Tone and a Short Glide Tone with a Common Gap Crossing Each Other
<5> A Long Glide Tone and a Short Glide Tone Crossing Each Other in Opposite Phases

II. The Split-Off Illusion

<6> The Split-Off Illusion in Glides in Different Directions
<7> The Split-Off Illusion in Glides in the same Direction
<8> The Split-Off Illusion with an Additional Glide
<9> The Split-Off Illusion in a Physically Bouncing Components
<10> The Split-Off Illusion in Repetition

III. Abstraction of Musical Melodies

<11> Successive Piano Tones
<12> Successive Piano Tones with the Sustaining Pedal
<13> A Melody of Silences

IV. The Illusory Continuity

<14> The Illusory Continuity of a Steady-State Tone
<15> The Illusory Continuity of a Glide Tone
<16> The Illusory Continuity of a Glide Tone with a Frequency Jump
<17> A Backward Extension of a Tone
<18> Music Sound Restoration

V. Time Perception

<19> Overestimation of a Divided Time Interval
<20> Time-shrinking
<21> A Discontinuous Change of Time Perception Caused by Time-shrinking
<22> Time-shrinking in Patterns Consisting of Three Empty Time Intervals
 
 

back intro forward
Introduction

We have made these auditory demonstrations to help understanding our research on auditory organization and time perception. Most of them were originally prepared for our research group, but we thought it might be convenient to make them accessible to other researchers. Unfortunately, we are not very much skilled in this kind of business, and, therefore, we cannot offer attractive animations. We simply offer some wave files for Windows 95.

We will improve the situation in the future (but not in the near future). We hope the demonstrations are still of some interest to some people in the field of auditory perception. Each demonstration is accompanied by a program written in 'J' for Windows 95/NT, which is available from Iverson Software Inc. Yoshitaka Nakajima wrote a set of programs in this language, and this system 'MeloDemoHome' enables us to generate various auditory stimuli. An advantage of 'MeloDemoHome' is that it requires the users only a week or so to learn its basic function, and the programs to generate sounds are very short. A sample program 'MDHDEMO' is available here for anyone interested in 'MeloDemoHome'. In order to run the accompanying programs to generate the sound patterns in the present demonstrations, 'J' and 'MeloDemoHome' are necessary.

 
 
 

I. The Gap Transfer Illusion


back play forward
<1> A Long Glide Tone and a Short Glide Tone Crossing Each Other
A long ascending glide tone and a short descending glide tone cross each other. The ascending tone is 2500 ms and from 422 to 2371 Hz, and the descending tone is 500 ms and from 1189 to 841 Hz. They cross each other in the middle of the pattern at 1000 Hz. The rate of frequency changes is always 1 oct./s. The rise time and the fall time are 3 ms except for the beginning and the end of the whole stimulus pattern, where they are 500 ms. A percept of 'bouncing' is common in patterns where two components cross each other in the same speed. However, the present pattern rarely gives such a percept. One of the possible percepts is a long ascending tone and a short descending tone. Quite often, however, the percept of the long component shows a sigmoidal curve.

Fig.01




back play forward
<2> A Long Continuous Glide Tone and a Short Disrupted Glide Tone Crossing Each Other
This pattern is almost the same as the pattern in Demonstration <1>. The difference is that the short descending glide tone has a 100-ms temporal gap in the middle. This pattern is usually perceived veridically.

Fig.02




back play forward
<3> The Gap Transfer Illusion
In this demonstration, most of the stimulus features are the same as those in Demonstration <1>. The difference is that the long ascending glide has a 100-ms temporal gap in the middle. The two glides are perceived as crossing each other, but the gap is perceived in the middle of the short descending tone. This percept is quite similar to that in Demonstration <2> despite the difference between the stimulus patterns.

Fig.03




back play forward
<4> A Long Glide Tone and a Short Glide Tone with a Common Gap Crossing Each Other
In this demonstration, two crossing glides with the same features as the glides in Demonstration <1> are presented, but both the long ascending glide and the short descending glide have a temporal gap of 50 ms at the crossing point in the middle. The listener typically perceives a continuously ascending tone and a short tone with a temporal gap in the middle.

Fig.04




back play forward
<5> A Long Tone and a Short Glide Tone Crossing Each Other in Opposite Phases
This stimulus pattern is the same as that in Demonstration <1> except for the phase relationship between the two glide tones. The two glide tones are in opposite phases at the crossing point, whereas they are in the same phase in Demonstration <1>. The present phase relationship yields a small gap or dip in the temporal envelope at the crossing point. The short glide tone is typically perceived as discontinuous, while the long one as continuously ascending.

Fig.05

 
 
 

II. The Split-off Illusion


back play forward
<6> The Split-off Illusion in Glides in Different Directions
An ascending tone glide of 1000 ms from 500 to 1000 Hz and a descending glide of 1000 ms from 800 to 400 Hz are presented successively with a temporal overlap of 200 ms in the middle. Thus, the whole duration of the pattern is 1800 ms. The rise time and the fall time are 5 ms. The listener typically perceives one long tone, which rises and then falls, and a short tone in the middle. The onset of the descending glide tone and the termination of the ascending glide tone forms a new tone perceptually.

Fig.06




back play forward
<7> The Split-off Illusion in Glides in the Same Direction
This pattern consists of two successive glides with a short overlap in time. The first glide is 1000 ms and from 500 to 1000 Hz. The second one is also 1000 ms and from 1306 to 2612 Hz. They have an overlap of 200 ms, thus making the whole duration of the pattern 1800 ms. The rise time and the fall time are always 5 ms. Typically, the listener perceives one long ascending glide tone through the whole duration of the pattern, and splits off a short tone in the middle perceptually.

Fig.07




back play forward
<8> The Split-off Illusion with an Additional Glide
In order to mask possible combination tones in the pattern of the split-off illusion, a masking glide tone was added to the pattern of Demonstration <7>. The masking glide tone is 1800 ms and from 250 to 871 Hz, always giving the fundamental of the whole complex tone. The rise time and the fall time are always 5 ms. The typical percept is almost the same as that in Demonstration <7>.

Fig.08




back play forward
<9> The Split-off Illusion in a Physically Bouncing Components
This pattern was made by separating the pattern of Demonstration <1> into the lower and the higher components and shifting them in time and in frequency. The lower component is a 1300-ms ascending glide tone from 380 to 933 Hz immediately followed by a 100-ms descending glide tone from 933 to 871 Hz. The higher component starts 1000 ms after the beginning of the first component. This component is a 100-ms descending glide tone from 1148 to 1072 Hz immediately followed by a 1300-ms ascending glide tone from 1072 to 2630 Hz. Thus, the difference between the two turning points is 0.2 octave in frequency and 100 ms in time. The rise time and the fall time are 500 ms for both ends of the pattern and 3 ms in the other places. The listener will hear a long continuously ascending tone and a short continuously descending tone. Even though the physical components are 'bouncing,' the perceptual components are 'crossing.' While it is well known that physically crossing pattern can yield the percept of 'bouncing,' the present pattern gives a reverse example. After the whole pattern is presented, each of the two turning components is presented separately.

Fig.09




back play forward
<10> The Split-off Illusion in Repetition
These demonstrations show variations of the split-off illusion. An ascending pattern and a descending pattern are alternately presented without silence. In the first pattern, an ascending glide of 1200 ms from 500 to 2000 Hz and a descending glide tone of the same duration from 1500 to 375 Hz are presented in alternation with overlaps of 200 ms. The listener perceives a continuously rising and falling tone and alternating high and low short tones. In the second pattern, an ascending part consists of a glide tone of 600 ms from 500 to 1000 Hz and another glide tone of the same duration from 1000 to 2000 Hz with an overlap of 200 ms in the middle. The listener perceives a continuous tone which is rising and falling repeatedly and short tones in the middle of each rising or falling part.

Fig.10-1
Fig.10-2

 
 
 

III. Abstraction of Musical Melodies


back play forward
<11> Successive Piano Tones
A musical scale is played on the piano with short overlaps between the tones. Despite the physical overlaps, the listener will hear a coherent musical scale.




back play forward
<12> Successive Piano Tones with the Sustaining Pedal
A musical scale is played on the piano using the sustaining pedal. Decaying piano tones are simply added up. Still the listener hears a coherent musical scale.




back play forward
<13> A Melody of Silences
In continuously sounding seven tones, C, F, G, A, Bb,C, and D, short temporal gaps make a familiar melody. This seems a reversal of figure (sounds) and ground (silences). The listener will hear a clear melody picking up not the onsets of the silences but the terminations of the silences (the onsets of the tones).

Fig.13

 
 
 
 

IV. The Illusory Continuity


back play forward
<14> The Illusory Continuity of a Steady-State Tone
This demonstration consists of four successive patterns. The first pattern gives a typical example of the continuity effect. A 2550-ms steady-state tone of 800 Hz with a 150-ms temporal gap in the middle is presented. A 150-ms noise with a frequency range of 400 to 1600 Hz (two octaves) is inserted in the temporal gap. The level of the tone is 5 dB lower than that of the noise. The rise time and the fall time at both ends of the pattern are 200 ms. They are in the other places in the pattern they are 3 ms. The listener will hear the tone as continuous even though the tone is replaced by the noise at the middle part. In the second pattern, the noise is replaced by a silence. The third pattern is the same as the first one except that the noise has a frequency gap ranged from 566 to 1131 Hz (one octave). The listener will hear the tone as separated in the middle if he/she listens to it carefully. The fourth pattern is identical with the second one physically.

Fig.14




back play forward
<15> The Illusory Continuity of a Glide tone
This demonstration also consists of four successive patterns. The first pattern gives another example of the continuity effect. A 2550-ms glide tone from 200 to 3200 Hz with a 150-ms temporal gap in the middle is presented. The rate of the frequency change is constant in the logarithmic scale. A 150-ms noise with a frequency range of 400 to 1600 Hz (two octaves) is inserted in the temporal gap. The first part of the glide tone ends at about 650 Hz and the second part begins at about 985 Hz. The level of the glide tone is 5 dB lower than that of the noise. The rise and the fall time at both ends of the pattern are 200 ms. They are 3 ms in the other places. The listener will hear the glide tone as continuous even though the tone is replaced by the noise at the middle part. In the second pattern, the noise is replaced by a silence. The third pattern is the same as the first one except that the noise has a frequency gap ranged from 566 to 1131 Hz (one octave). The listener will hear the glide tone as separated in the middle if he/she listens to it carefully. The fourth pattern is identical with the second one physically.

Fig.15




back play forward
<16> The Illusory Continuity of a Glide Tone with a Frequency Jump
This demonstration is a variation of Demonstration <15>. A jump of frequency was introduced to the glide tone, dividing it clearly into two successive tone glides. The first glide is from 200 to 566 Hz, and the second from 1131 to 3200 Hz. In the third pattern, the first glide ends at the lower boundary of the frequency gap of the noise, and the second glide starts from the upper boundary. In the third pattern, the listener is able to hear a stronger continuity than in Demonstration <15> even though the principle of good continuity must work to a smaller degree. The clearer clues of the termination and the onset seem to destroy the continuity of the glide tone in Demonstration <15>. On the other hand, the termination and the onset of the glide tones in the middle of the present pattern are obscured by the noise.

Fig.16




back play forward
<17> A Backward Extension of a Tone
Two pure tones of 800 Hz and 200 ms are presented successively, and the first one is preceded immediately by a noise of 1000 ms and 400-1600 Hz. The level of the noise is 12 dB higher than that of the tones. The rise time and the fall time of the pure tones are 6 ms, and those for the noise are 200 ms and 3 ms respectively. These pure tones are presented again without the noise. The duration of the pure tone preceded by the noise tends to be overestimated.

Fig.17




back play forward
<18> Music Sound Restoration
One tone of a melody from "Turkish March" (W. A. Mozart) is replaced by a white noise. Even though one note is deleted, we can still hear a complete melody. Our auditory system restores the missing sound in accordance with the melodic context. The melody and the noise are segregated from each other perceptually, which makes it difficult to judge the temporal position of the noise.

 
 
 
 

V. Time Perception


back play forward
<19> Overestimation of a Divided Time Interval
When an empty duration of 240 ms marked by two short tone bursts is divided into two empty parts by another tone burst, the whole duration is often overestimated. However, a clear overestimation appears only when the listener perceives the divided duration as divided. A divided duration and a non-divided (empty) duration are presented alternately.

Fig.19




back play forward
<20> Time-shrinking
A short empty duration immediately preceded by an even shorter empty duration can be underestimated. A fixed empty duration of 240 ms is immediately preceded by another empty duration of 80-320 ms. In the first presentation, the preceding duration is 320 ms, and it is decreased in steps of 20 ms in the following presentations. The second duration is perceived as if it were shortened when the preceding duration reached around 180 ms (the eighth presentation).

Fig.20




back play forward
<21> A Discontinuous Change of Time Perception Caused by Time-shrinking
In the first presentation of the first series, three sound markers define two empty durations of 160 ms each. The first duration is decreased in steps of 10 ms, and the second duration is increased in steps of 10 ms. Thus, the total duration is fixed at 320 ms. When the difference between the first and the second duration is up to 80 ms (the fifth pattern), 'time-shrinking', i.e., the underestimation of the second duration caused by the presence of the first duration takes place, and the two neighboring durations are perceived as very similar to each other. When the physical difference between these durations is enlarged further, time-shrinking disappears, and the difference between the neighboring durations is suddenly perceived clearly.
This kind of sudden change of percept does not take place when the temporal order between the two durations is reversed as in the second series.

Fig.21




back play forward
<22> Time-shrinking in Patterns Consisting of Three Empty Time Intervals
In the first presentation (the control condition), two isolated empty durations of 240 ms are perceived as almost equal to each other. In all the other presentations, three neighboring empty durations, among which the second and the third are fixed at 160 and 240 ms, are followed by an isolated empty duration of 240 ms. The first empty duration is 40 ms in the second presentation, and increased in steps of 40 ms in the nine following presentations. The third empty duration is underestimated when the first empty duration is 160 ms (the fifth presentation), 320 ms (the ninth presentation), or above.

Fig.22-1
Fig.22-2
Fig.22-3
Fig.22-4
Fig.22-5
Fig.22-6
Fig.22-7
Fig.22-8
Fig.22-9
Fig.22-10
Fig.22-11




Contents

Demo Index

Top Page