The tritone paradox and the Simon effect : a study of pitch perception

Transcription

1 Honors Theses Psychology Spring 2017 The tritone paradox and the Simon effect : a study of pitch perception Olivia Anne Coackley Whitman College Josie Nikita Furbershaw Whitman College Samantha Anne Fata Whitman College Penrose Library, Whitman College Permanent URL: This thesis has been deposited to Whitman College by the author(s) as part of their degree program. All rights are retained by the author(s) and they are responsible for the content.

2 The Tritone Paradox and the Simon Effect: A Study of Pitch Perception by Olivia Anne Coackley, Samantha Anne Fata, and Josie Nikita Furbershaw A thesis submitted in partial fulfillment of the requirements for graduation with Honors in Psychology. Whitman College 2017

3 Certificate of Approval This is to certify that the accompanying thesis by Olivia Anne Coackley has been accepted in partial fulfillment of the requirements for graduation with Honors in Psychology. Walter Herbranson Whitman College May 10, 2017 ii

4 Table of Contents Abstract... v List of Tables and Figures... vi Introduction... 1 Ambiguity in the Perception of Pitch... 1 Ambiguity in the Perception of Visual Figures... 4 Manipulating the Perception of the Tritone... 6 The Simon Task... 7 Hypotheses... 9 Method Participants Design Stimuli Procedure Measures Results Main Hypotheses Unplanned analyses Discussion Hypothesis Hypothesis Hypothesis Hypothesis iii

5 Strengths, Limitations, and Future Directions Final Thoughts References Tables Figures Appendices iv

6 Abstract The tritone paradox is a phenomenon of auditory perception in which individuals perceive octave-ambiguous tritone intervals differently from other individuals. Although they are by definition ambiguous, it is unclear whether perception of these tritones can be manipulated. The purpose of this study was to determine if the perception of ambiguous tone pairs (tritones) is susceptible to priming, similar to ambiguous images. Participants listened to a series of tritone pairs and non-tritone pairs in the context of a Simon task and judged if the tones were ascending or descending. The results demonstrated that under some circumstances, perception of tritones can be primed. This supports a model of pitch perception that includes influence of external cues. Future research should focus on the difference between perceived and actual direction of primes, the possibility that some non-tritone intervals may also be ambiguous, and the importance of including participants from a wide range of linguistic and geographical backgrounds. Keywords: auditory perception, cognitive psychology, octave-ambiguous tones, priming, Simon effect, stimulus ambiguity, tritone paradox v

7 List of Tables and Figures Table 1. Participant Demographic Information Table 2. Hypothesis 1: 2x2 ANOVA on Response Times to Standard Tone Pairs Table 3. Hypothesis 1: 2x2 ANOVA on Accuracy of Standard Tone Pairs Table 4. Hypothesis 3: 2x2 ANOVA on Response Times for Tritones Table 5. Hypothesis 2: Descriptive Statistics and Results of One-Sample t-tests of Tritone Perception during Baseline Table 6. Hypothesis 2: Descriptive Statistics and Results of One-Sample t-tests Comparing Tritone Consistency and Standard Tone Accuracy Table 7. Hypothesis 4: Descriptive Statistics and Results of One-Sample t-tests Comparing Tritone Priming to Baseline Perception Table 8. Hypothesis 2: Descriptive Statistics and Results of One-Sample t-tests Comparing Tritone Perception in the Baseline Between Participants from the West Coast and Rest of the Country Figure 1. Rat-man image Figure 2. Rising pitch helix Figure 3. Pitch class circle Figure 4. Proximity Figure 5. Differently oriented pitch class circles Figure 6. Hypothesis 1: Mean reaction time to standard tone stimuli Figure 7. Hypothesis 1: Mean accuracy rate to standard tone stimuli Figure 8. Hypothesis 3: Mean reaction time to tritone stimuli Figure 9. Example pitch class circles vi

8 Figure 10. Percentage of tones heard as ascending as a function of pitch class Figure 11. Percentage of tones heard as descending as a function of pitch class (Deutsch) vii

9 The Tritone Paradox and the Simon Effect: A Study of Pitch Perception A fundamental principle of perception is that it is largely subjective. One compelling line of evidence for this proposition is the existence of perceptual illusions and ambiguous figures that can be demonstrated in any sensory modality. For example, Bugelski and Alampay (1961) presented ambiguous visual images that could be interpreted in two different ways: as either a rat or a man, as shown in Figure 1. Whether participants saw the image as a rat or as a man depended on whether they had previously been shown unambiguous images of animals or of people. While many researchers have studied ambiguous visual figures and their properties, relatively little research has been completed regarding ambiguous auditory stimuli. Consequently, some of the details of auditory perception are not as well understood. For example, it is currently unclear whether pitch perception is dependent on context and external cues or on a rigid internal understanding of pitch that is not easily swayed by context. One potential auditory equivalent to reversible visual images such as the rat/man figure is a tritone. The tritone, or interval of exactly half an octave, provides a potentially useful tool for the study of pitch perception because tritones can be heard as either ascending or descending when specific aural cues are removed. The capability, or lack thereof, to manipulate the perception of a tritone has important implications for current theories of pitch perception, as well as for the study of ambiguous auditory stimuli. Ambiguity in the Perception of Pitch The perception of normal, unambiguous tones depends primarily on two dimensions of pitch: height and class. The dimension of height is a continuum that describes a pitch s relative position as high or low and physically corresponds to the

10 frequency of sound waves. Height allows a listener to distinguish between octaves. In contrast, pitch class is a circular dimension that defines a pitch s position within an octave. Each note within a pitch class (e.g., pitch class C#) can be considered perceptually similar, regardless of its octave. This similarity is due to the proportional relationship between the fundamental frequencies of notes in the same pitch class. Together, these two dimensions of pitch perception can be represented as a rising helix, completing one full turn each octave, as illustrated in Figure 2. Each instance of a particular pitch class is vertically aligned with every other instance of that same pitch class within the helix. If aural cues for pitch height are removed, the helix collapses into a circle, and pitch perception becomes entirely dependent on pitch class (and not pitch height), as illustrated in Figure 3. Stimuli that have a well-defined pitch class but cannot be linked to a specific octave are known as Shepard tones or octave-ambiguous tones. These tones are created by combining pure-tone components of the same pitch class from different octaves (e.g., one tone consists of C 1, C 2, C 3, and so on, while another consists of C# 1, C# 2, C# 3, and so on) with amplitudes arranged in a bell-shaped envelope so that the tones in the middle of the musical range are louder than tones in higher and lower registers (Deutsch, 1997). Because Shepard tones do not consist of information from one specific octave, the listener cannot use pitch height to discriminate between two Shepard tones played in sequence. Instead, the listener must rely entirely on the differences between two pitch classes. This discrimination is made based on proximity, which describes how close notes are to each other on the pitch class circle (see Figure 3). Generally, listeners preference the shortest possible distance between two tones on 2

11 the pitch class circle in making such decisions (Shepard, 1964). For example, the interval C-E would more often be interpreted as an ascending interval of four halfsteps, or semitones (C, C#, D, D#, E), than a descending interval of eight semitones (C, B, A#, A, G#, G, F#, F, E), as demonstrated by Figure 4. However, when Shepard (1964) presented pairs of pitches that were separated by exactly a half octave ( tritone ), proximity was no longer a useful cue as both paths around the pitch class circle consisted of the same number of steps. The interval B-F, for example, could either be considered an ascending interval of six semitones (B, C, C#, D, D#, E, F) or a descending interval of six semitones (B, A#, A, G#, G, F#, F). Within a large sample, this interval was judged as ascending or descending with equal frequency, similar to how an ambiguous cube can be seen in either of two orientations with equal frequency (Howard, 2012). The ambiguity of this half-octave interval is the basis for what is known as the tritone paradox. For an ambiguous cube and other visual stimuli such as the rat-man figure (Bugelski & Alampay, 1961), viewers are able to switch between different interpretations at will once they become aware of both of the interpretations. However, unlike these visual stimuli, perception of the tritone paradox appears to vary from person to person, rather than from viewing to viewing. While participants may be able to force themselves to perceive an ambiguous cube a different way, participants may be predisposed to consistently interpret tritones as moving in a specific direction regardless of their awareness of the tritone paradox. The tritone paradox occurs because most people are biased to hear some pitch classes as higher than others. That is, Shepard tones in one region of the pitch class circle are generally perceived as 3

12 higher than tones in the opposite region of the circle. In this way, we might think of different individuals as having a different orientation of their pitch class circles, as represented in Figure 5. As a result, most listeners consistently perceive certain tritones as rising (or ascending) and others as falling (or descending). Deutsch (1987) first demonstrated the tritone paradox in the general population as a systematic influence of pitch class on the perceived height of a pitch, meaning that certain pitch classes were consistently perceived as higher than others. However, individual participants disagreed on the perceived heights of these pitch classes. More specifically, participants disagreed on whether or not specific tritones were ascending or descending. For example, one participant may have very confidently perceived the interval B-F as descending (left panel in Figure 5), another participant may have very confidently perceived the same interval as ascending (center panel), and a third might have been uncertain (right panel). This disagreement is due to widespread variation in the orientation of the pitch class circle across participants. The orientation of the pitch class circle is determined by a wide range of factors, including the range of the speaking voice (Deutsch, North, & Ray, 1990), hometown and language spoken (Deutsch, 1991), and speech patterns heard early in life (Deutsch, Henthorn, & Dolson, 2004). Musical training and experience does not seem to have an effect on the presence of the tritone paradox (Deutsch, Kuyper, & Fisher, 1987). Ambiguity in the Perception of Visual Figures Because octave-ambiguous tritones are ambiguous auditory stimuli, it may be possible to manipulate their perception, much as one can influence the perception of ambiguous visual stimuli. One method of manipulating perception of an ambiguous 4

13 visual stimulus is priming. Priming is the effect in which exposure to a specific stimulus either facilitates or interferes with the processing of a subsequent stimulus. Researchers have studied the effect of priming on ambiguous visual figures extensively. In a pioneering study, Gopnik and Rosati (2001) found that children older than five were able to switch between two interpretations of an ambiguous image, whereas younger children were not able to do so, even when the second interpretation was pointed out to them. Bugelski and Alampay (1961) were able to influence participants perceptions of the rat/man image by presenting several non-ambiguous images of animals or humans before the ambiguous image. Scocchia, Valsecchi, Gegenfurtner, and Triesch (2013) showed that participants perceptions of ambiguously rotating spheres could be primed by images held in working memory. The researchers instructed participants to hold an unambiguous image of a rotating sphere in their working memory while an ambiguously rotating sphere was viewed on a computer screen. The unambiguous sphere in working memory influenced the perceived direction of rotation of the ambiguous sphere on the screen. Such priming effects, however, may not be universal. In a study motivated by failed attempts to replicate priming of ambiguous figures, Wilton (1985) found two mediating factors: the duration of the prime preceding the ambiguous stimulus and the interval of time between the prime and the ambiguous stimulus. These results indicate that priming effects on the interpretation of ambiguous figures are not consistently robust and depend on the implementation of these two factors. 5

14 Manipulating the Perception of the Tritone In the specific case of the tritone paradox, it is unclear whether priming can change an individual s perception of a tritone. In a series of experiments examining the effects of preceding tone pairs on the perceived direction of ambiguous tone pairs, Giangrand, Tuller, and Kelso (2003) found that a series of pitches were perceived as ascending or descending, not based on frequency (Hz), but based on the direction of the preceding tone pairs. That is, if the preceding tone pair was ascending, then participants perceived the ambiguous tone pair as ascending, but if the preceding tone pair was descending, then the participants perceived the ambiguous tone pair as descending. In contrast, Repp and Thompson (2010) did not find any significant priming effects when octave-ambiguous tritones (consisting of pairs of Shepard tones) were preceded by unambiguous tritones (that had a clearly defined octave in addition to pitch class). These two results lend themselves to competing theories of pitch perception. The first holds that pitch perception relies on external context and cues, such as the Gestalt principle of good continuation. If this theory is correct, perception of ambiguous auditory stimuli should be similar to the perception of ambiguous visual stimuli, and therefore should be susceptible to priming. The contrasting theory holds that these ambiguous tones are understood based on internal cues that are not easily swayed by context, and thus not susceptible to priming. If the perception of ambiguous tone pairs is similar to the perception of an ambiguous image, it should be possible to manipulate perception of tritones through the influence of a preceding tone pair. On the other hand, if perception of ambiguous tone pairs is based on internal rather than external cues, this perception should be significantly more difficult to manipulate. 6

15 Priming is, of course, only one way to influence a response. In addition to prior events, responses can also be facilitated or inhibited by concurrent factors, such as the location of the relevant response on a display interface. Examples of this kind of influence include the Stroop effect and the Simon effect. The Simon Task When designing the spatial arrangement of an interface, intuitive layouts are preferential to non-intuitive layouts. The intuitiveness of an interface s spatial arrangement is referred to as stimulus-response compatibility (S-R compatibility; Fitts & Seeger, 1953). Therefore, an interface that has high S-R compatibility is more intuitively organized than an interface with low S-R compatibility. Manipulating an interface s spatial arrangement so that responses are placed in atypical or non-intuitive locations is the basis of several experimental procedures. In the Simon task, for example, a participant hears an audio stimulus monaurally through headphones. The participant is then asked to respond based on the stimulus that was heard, without regard for which ear the stimulus was heard in. For example, participants respond by pressing a key with their left hand after hearing the word left in either ear, and by pressing a key with their right hand after hearing the word right in either ear. A trial is considered congruent when the stimulus is presented on the same side as the required response (e.g., a stimulus presented to the right ear leads to a response with the right hand), and thus has a high S-R compatibility. Likewise, a trial is considered incongruent when the stimulus is presented on the side opposite the required response (e.g., a stimulus presented to the left ear leads to a response with the right hand), and thus has a low S-R compatibility (Simon & Rudell, 1967). 7

16 When the S-R compatibility of a trial is low (on incongruent trials), participants take longer to respond and make more errors than when S-R compatibility is high (on congruent trials; Simon & Rudell, 1967). Note that in this case, an irrelevant aspect of the task (response congruency) has an effect on the way information is processed. The Simon task could therefore be a way to investigate additional factors used to disambiguate ambiguous stimuli. Additionally, the Simon task provides a kind of measure of confidence, as it may demonstrate whether participants perceive ambiguous stimuli as being ambiguous. There is minimal prior research into how the Simon effect is influenced by ambiguous stimuli, visual or otherwise. It stands to reason that if participants perceive the stimulus as being ambiguous and have difficulty deciding which interpretation of the figure is correct, the expected pattern of high accuracy and short response times on congruent trials with low accuracy and long response times on incongruent trials will not be demonstrated. However, if participants perceive a specific interpretation confidently (without spontaneously perceiving the other interpretation), this pattern may still be preserved. This can be referred to as an illusory Simon effect. Because ambiguous stimuli, such as the octave-ambiguous tritone, have more than one correct interpretation, this illusory Simon effect is based primarily on whether the individual s perception of the ambiguous stimulus is congruent or incongruent with the side of the body the response requires. This effect may only be demonstrated through longer or shorter response times because there is no correct answer to form the basis of a measure of accuracy. The possibility of demonstrating an illusory Simon effect, and 8

17 therefore similarity in the perception of non-ambiguous and ambiguous stimuli, makes the Simon task an incredibly valuable tool in the study of ambiguous figures. Hypotheses The goal of this study was to see if perception of tritones as ascending or descending could be manipulated. Additionally, this study sought to investigate whether or not these tritones are perceived confidently or are perceived as being ambiguous. Based on previous research regarding the Simon task, the tritone paradox, and perceptual priming, we expected to confirm two existing effects in a new context: 1. Congruent trials should produce shorter RTs and fewer errors than incongruent trials (the Simon effect ); and 2. Not all participants should hear the same tritone pairs in the same way (as ascending and descending), but each participant should perceive certain tritones as consistently ascending and others as consistently descending (the tritone paradox ). In addition, we expected two new findings: 1. Ambiguous tritones perceived as congruent (e.g., presented to the right ear and resulting in a right-hand response) should have shorter RTs than tritones that are perceived as incongruent, demonstrating an illusory Simon effect (based on participants disambiguation of tritone stimuli, rather than on the stimuli themselves); and 2. Tritones should be more likely to be perceived the same way as the actual direction of the immediately preceding unambiguous tone pair, demonstrating a priming effect on the perception of the tritones. 9

18 Method Participants Participants were recruited through s to listservs, posters around campus, and by word of mouth. We used convenience sampling to maximize the number of participants. Participants were excluded if they were not a student at Whitman College, or if they had hearing impairments. No participants withdrew from our study. This procedure yielded a sample of 67 college students from Whitman College, a small liberal arts college in Walla Walla, Washington. Five participants were excluded from the study because they identified less than 60% of the standard tone pairs correctly. Our final sample consisted of 62 participants between the ages of 18 and 24 (M = , SD = 1.334). This included 20 men, 41 women, and one person who did not indicate their sex. There were 21 first-year participants, 16 sophomores, nine juniors, and 16 seniors. Our sample included 55 right-handed participants. The ethnicity sample was made up of 47 participants who identified as White, 5 participants who identified as multiple ethnicities, 4 participants who identified as Asian, 3 participants who identified as Asian-American, 2 participants who identified as Hispanic, and 1 participant who identified as Indian. Almost half of our participants were from the Pacific Northwest (either Washington or Oregon, n = 30). Other participants were from California (n = 7) or from other states not on the West Coast (n = 18). Several participants were excluded from geographic analyses because they grew up in multiple locations (n = 4) or were born outside of the United States (n = 2). Previous musical experience, or experience playing an instrument or with voice or choir, was identified in 59 participants. These participants displayed a median of eight 10

19 years of music experience (IQR = 5-12). Participants played a variety of instruments with the most common being piano (n = 38), voice (n = 30), guitar (n = 14), and violin (n = 11). Two participants reported that they possessed absolute pitch (the ability to name a pitch without a reference pitch). Bilingualism or multilingualism was selfreported by 23 participants (we excluded a participant who spoke American Sign Language in multilingual analyses due to the fact that it is not a spoken language). Additional demographic information can be found in Table 1. Design The experiment was a 3x2 within groups factorial experiment with tone pair type (ascending, descending, or tritone), and ear of tone presentation (right or left) as independent variables. The two primary dependent variables were reaction time (measured in s) and accuracy (percentage of unambiguous Shepard tone pairs correctly identified). In addition, for the subset of trials involving tritones, perception (of the tritone as ascending or descending) and consistency (degree to which perception of specific tritones remained constant across repeated presentations) were measured. Stimuli All stimuli used in the following measures were Shepard tones (octaveambiguous tones), which made it impossible to determine which octave each note was from, though the pitch class of each note (C, C#, etc.) could be easily discerned. Tone pairs were either unambiguous standard tone pairs (less than six semitones apart) or tritones (exactly six semitones apart). Tones in tone pairs were presented for 500 ms each, with no delay between them. Thus, the presentation of each pair lasted 1000 ms. This presentation of tones emulated the procedure used previously by Deutsch (1991). 11

20 Tritone pairs were from Deutsch s (1995) CD, and standard tone pairs were cut from the tritone pairs and recombined in Amadeus Pro (HairerSoft: Kenilworth UK). Both standard and tritone pairs were then edited in Garageband (Apple: Cupertino, CA) to play in either the left or right auditory channel. Reaction time to all tone pairs was measured by Qualtrics as the time between when the page loaded (the tone played as soon as the page loaded) and a key being selected. Procedure This study was approved by the Whitman College Institutional Review Board. This study took place in an on-campus computer laboratory in order to minimize both visual and auditory distractions and control for the environment. Participants all received the same information through standardized instructions on Qualtrics. Participants either used provided earbuds or brought in their own earbuds or headphones for the study. Participants first completed a tritone baseline measure to establish a measure of each participant s tritone perception. Participants then completed an accuracy check to make sure that they were able to correctly identify ascending and descending standard tone pairs. Next, participants completed a Simon task practice trial in order to confirm that participants understood the Simon task. Finally, participants took part in the Simon task with tritones measure. Measures Tritone baseline. Participants listened to a series of trials that allowed us to determine their baseline perception of tritones. All participants heard the same sequence of tritones, with each of the 12 possible tritones played three times each (for a 12

21 total of 36 presentations). Tones were presented to both ears. Participants were told to determine whether the tone pair was ascending or descending and respond on the keyboard by pressing the m or c keys, respectively. Participants were encouraged to respond as quickly as possible but were given no time limit for a response. The tritone baseline took about 3 minutes and 36 seconds. See Appendix A for the full list of tone pairs and the specific sequence used. Accuracy check. Participants listened to a series of unambiguous pairs (differing by less than six semitones) of Shepard tones to confirm that they could make accurate judgments of standard ascending and descending Shepard tones. Five ascending tone pairs were played and five descending tone pairs were played in the same order for all participants. Tones were presented to both ears. Participants were required to repeat the procedure up to three times until they identified the directionality of nine of the ten tone pairs correctly. This portion of the study took about 60 seconds. See Appendix B for a full list of tone pairs and the specific sequence used. Simon task practice trial. Participants were given an opportunity to practice the Simon task. All participants heard the same eight standard (non-tritone) tone pairs in the same order and in the same ears. Tone pairs were presented monaurally, and the order in which tones were presented to each ear was randomized. Participants again were instructed to determine if each tone pair was ascending or descending and respond on the keyboard by pressing either the m key or the c key. This portion of the experiment took about 48 seconds. Participants were required to repeat the procedure up to three times until they were able to identify the directionality of five of 13

22 the eight tone pairs correctly. See Appendix C for a full list of the tone pairs, the specific sequence used, and the ears in which tones were played. Simon task with tritones. Participants completed four blocks of trials. Each block consisted of 84 tone pairs: 24 tritones (2 of each tritone) and 60 standard tone pairs. Half of these 60 standard tone pairs were ascending and the other half were descending. Additionally, each standard tone pair was presented in an ascending sequence twice and in a descending sequence twice. Furthermore, since each tritone appeared twice, one instance was preceded by an ascending tone pair and the other was preceded by a descending tone pair. The number of tone pairs between tritones was randomized. Tone pairs were presented monaurally, and the order in which tones were presented to each ear was randomized. After every 21 tone pairs, there was an optional 20 second break, for a total of three optional 20 second breaks. Each block lasted about five minutes. Participants then had an optional one-minute rest period between each block. See Appendix D for the full order of the tone pairs in each block. Follow up. After the fourth and final block of the Simon task with tritones measure, participants filled out an online form requesting demographic information (see results in Table 1). Within seven days, participants received a link to the initial Tritones baseline measure via and were encouraged to complete it at their leisure. Before starting the experiment, all participants gave informed consent after the experimenter explained the voluntary nature of participation and the pertinent risks. Participants who completed the study received a cookie or an orange as incentive. Participants who completed the at-home portion of the study were entered into a raffle for a gift card. Participants names only appeared in two locations: on the consent 14

23 form and on a separate spreadsheet of participant s names and assigned identification numbers. The rest of the information, including responses to tone pairs and demographic information, was stored by identification number on Qualtrics and was password-protected. Participants names could not later be connected back to their results because a name was never associated with a participant s results. Results Main Hypotheses Hypothesis 1: Congruent trials will have shorter response times and fewer errors than incongruent trials. A 2 (ear: left, right) by 2 (hand: left, right) withinsubjects Analysis of Variance (ANOVA) was conducted on response times to nontritone stimuli (see Table 2). There was no significant effect of hand, F(1, 61) = 1.671, p =.201, η² =.027. Ear had a significant effect, F(1, 61) = 4.504, p =.038, η² =.069, with the right ear producing faster response times than the left ear. Lastly, the interaction between ear and hand was not significant, F(1, 61) = 2.492, p =.120, η² = (See Figure 6). Another 2 (ear: left, right) by 2 (hand: left, right) within-subjects ANOVA was conducted on accuracy of responding to non-tritone stimuli (see Table 3). Hand had no effect on accuracy, F(1, 61) =.935, p =.337, η² =.015. Ear had a significant effect on accuracy, F(1, 61) = , p =.001, η² = 0.179, with the right ear being more accurate than the left ear. The interaction between ear and hand was significant, F(1, 61) = , p <.001, η² = (See Figure 7). Hypothesis 3: Tritone trials perceived as congruent would produce shorter response times than tritone trials perceived as incongruent. A paired samples t-test 15

24 was conducted to compare the mean response times on congruent and incongruent trials. There was a significant difference between the response times of congruent (M = 1.733, SD = 0.469) and incongruent (M = 1.787, SD = 0.479) trials; t(61) = , p =.003; d =.114, such that participants were slower to respond to incongruent trials. To further explore this congruency effect, a 2 (ear: left, right) by 2 (hand: left, right) within-subjects ANOVA was conducted on response times (see Table 4). Results indicated that hand had a significant effect on response time, F (1, 60) = 6.461, p =.014, η² =.097, such that the left hand was faster than the right hand. Ear had no effect on response time, F(1, 60) = 2.342, p =.131, η² =.038. Lastly, the interaction between ear and hand had a significant effect on response time, F (1, 60) = 7.029, p =.010, η² = The nature of the interaction can be seen in Figure 8. Note that one participant was not included in this analysis because their responses did not vary sufficiently (they did not respond incongruently for any tone pairs played in their left ear). Hypothesis 2: Participants' responses will replicate the tritone paradox. There are several aspects of the tritone paradox. The first is that different people hear the same tritones differently (i.e., some people will hear a specific tritone as ascending while others will hear the same tritone as descending). In order to test this, we calculated the percentage of instances in which participants classified each of the 12 tritones as ascending. An average of around.5 would confirm this aspect of the hypothesis, indicating that equal numbers of participants perceived a tritone as ascending and as descending during the tritone baseline trials. We then ran a series of 12 single-sample t-tests (one for each tritone), each with a null hypothesis of.5 (see Table 5). Six out of the 12 tritones had means that were significantly different from.5, 16

25 while the other six tritones were not. The tritones that were not significantly different from.5 were A#-E, B-F, D-G#, A-D#, G-C#, and G#-D. The tritones that were significantly different from.5 were C-F#, C#-G, D#-A, E-A#, F-B, and F#-C. The means on Table 5 also indicate that five of the tritones were heard on average as ascending (M >.5) and seven of the tritones were heard on average as descending (M <.5), indicating no systematic bias toward perceiving tritones in either direction. Note also that four of the five tritones heard as ascending were adjacent to one another on the pitch class circle, with no intervening tritones that were heard as descending. Similarly, six of the seven heard as descending were adjacent to one another. The average percentage of instances in which tritones were perceived as ascending ranged from 28.0% (D#-A) to 64.0% (C-F#). Note that of the 12 tritones, only six are unique combinations of tones. The other six combinations consist of the same tones presented in reversed order (e.g., B and F are paired twice, once as B-F with B first and F second, and again as F-B with F first, and B second). Only one tritone pair was perceived the same way when presented in both orders: C#-G was heard as significantly descending, and G-C# was also heard as descending (though not significantly so). A second aspect of the tritone paradox is that individual participants have a characteristic pitch-class orientation, in which tritones identified as ascending and descending should be localized to separate regions of the pitch-class circle. This also means that certain tritone pairs should be more consistent for a participant and other tritone pairs should be more ambiguous (those consisting of tritones in the transitional area between regions perceived as higher and lower). We calculated average estimates 17

26 of consistency for each participant for each tritone, based on multiple presentations: a score of 1.0 meant that a participant responded to a tritone the same way every time it was presented. A score of 0.5 meant that a participant responded to the tritone as ascending and as descending equally often. We then conducted a series of one-sample t-tests, each with a null hypothesis of.748, which corresponded to standard tone pair accuracy in the main portion of the experiment. The goal of this test was to compare each participant s consistency in identifying tritones with their standard tone pair accuracy rates (assuming they reflect a ceiling for consistency). Eight tritones were significantly different from the average accuracy score: A#-E, C-F#, C#-G, D#-A, E- A#, F-B, G-C#, and G#-D (see Table 6). We also averaged each participant s accuracy score (M =.748, SD =.063) and averaged each participant s tritone consistency score (M =.807, SD =.078) to compare overall performance using one paired samples t-test, which indicated a significant difference between the two, t(61) = , p <.001; d =.832. Participants were, in general, more consistent for tritones than for non-tritone stimuli. We also recreated the pitch class circles of three participants, combining their baseline tritone results with their follow-up tritone results (for a total of six presentations of each tritone). These example pitch class circles can be found in Figure 9. A third part of this hypothesis was to confirm research that has already been conducted on the tritone paradox. Deutsch (1987) conducted a study in which she normalized the orientations of participants pitch class circles to see if, despite participants understanding specific pitches differently, participants all followed the same overall pattern with separate ascending and descending regions. We considered 18

27 each participant s responses to tritones individually and attempted to identify the first tritone that the participant consistently responded to as ascending, which was preceded by descending tritones and followed by other ascending tritones. This allowed us to normalize each participant s pitch class circle by selecting the first tritone that was consistently heard as ascending as the first tone pair of the sequence and then following this tritone with the rest of the tritones in chromatic order. We created averages for the first tritone in the sequence, then the second tritone in the sequence, and so on. Results of this analysis can be seen in Figure 10. Visual inspection shows that our graph is fairly similar to Deutsch s graph (reproduced as Figure 11 for comparison), with a continuous region of tone pairs generally perceived as ascending, another continuous region generally perceived as descending, and a transitional area between. Hypothesis 4: Tritones will be perceived in the same way as the immediately preceding unambiguous tone pair (a priming effect). We used two paired-samples t-tests to see if the presence of unambiguous primes affected how tritones were perceived. The mean percentage of successful ascending primes (tritones primed with ascending tone pairs and perceived as ascending; M =.403, SD =.162) was significantly different from the average percentage of tritones perceived as ascending during the baseline (M =.465, SD =.132), t(61) = , p =.004; d =.420. Tritones primed by ascending tone pairs were heard as ascending significantly less often than the tritones in the baseline, which was the opposite of our prediction. Our second paired-samples t-test compared the mean percentage of successful descending primes (tritones primed by descending tone pairs and perceived as descending; M =.558, SD =.153) to the average percentage of tritones perceived as descending during 19

28 the baseline (M =.535, SD =.132). The difference was in the expected direction (tritones primed by descending tone pairs were heard more often as descending than in the baseline), but this difference was not significant, t(61) = 1.193, p =.237; d =.161. The above analysis considers data from all 12 tritones together. To investigate the possibility that some tritones may be more easily primed than others, we repeated the above analysis treating each tritone separately. Of the 12 tritones, nine were significantly different from the baseline when primed with ascending tone pairs. Of these, two were perceived as ascending significantly more often than in the baseline: D#-A and F-B. The remaining seven tritones were perceived as descending significantly more often following priming with an ascending tone pair than in the baseline: A#-E, B-F, C-F#, D-G#, F#-C, G-C#, and G#-D. Of the 12 tritones primed with descending tone pairs, five tritones had primed mean percentages that were significantly different from the baseline percentage. Of these, three were perceived as descending significantly more often than in the baseline: C#-G, D#-A, and F-B. The remaining two tritones were perceived as ascending significantly more often than in the baseline: A-D# and C-F#. Full results can be seen in Table 7. Unplanned analyses Comparison of reaction times to standard tone pairs and to tritones. We compared the mean reaction times to non-ambiguous tone pairs and to tritones using a paired samples t-test. There was a significant difference between standard tone pairs (M = 1.994, SD = 0.421) and tritones (M = 1.801, SD = 0.516), such that participants responded on average more slowly to standard tone pairs than to tritones, t(60) = 5.074, p <.001; d =

29 Effects of right- and left-handedness. To investigate the effect of handedness on response time, we conducted two separate 2 (ear: left, right) by 2 (hand: left, right) within subjects ANOVAs, one for left-hand dominant individuals and one for righthand dominant individuals. There was not a significant difference between right- and left- handed participants and which hand was faster to respond with, F(1, 54) = 1.197, p =.279; η² =.022, F(1, 6) = 0.717, p =.429; η² =.107. There was, however, a significant effect of ear on left-handed individuals, such that tones played in the left ear had slower response times than tones played in the right ear, F(1, 6) = , p =.014; η² =.662. This difference was not found in right-handed individuals, F(1, 54) = 2.065, p =.157; η² =.037. There were no significant interaction effects for left-handed or right-handed participants, F(1, 54) = 2.565, p =.115; η² =.045, F(1, 6) = 0.010, p =.925; η² =.002. Hometown and tritone perception. We compared the average perception of each tritone for participants from the West Coast (California, Oregon, and Washington; n = 37) and participants from everywhere else in the United States (n = 18) to.5 using a series of one-sample t-tests to determine if hometown had an effect on certain tritones perceived ambiguity. Participants from the West Coast agreed that five of the tritones were either significantly ascending or significantly descending (D#-A, C-F#, F-B, E-A#, and F#-C). In comparison, participants from everywhere else in the United States agreed only on the direction of one tritone (D#-A; see Table 8). Seven participants were not included in this analysis because they reported multiple hometowns, grew up outside of the United States, or did not report a hometown. 21

30 Comparison of accuracy on ascending and descending standard tone pairs. We investigated whether accuracy rates differed for non-ambiguous ascending tone pairs and non-ambiguous descending tone pairs using a paired samples t-test. There was no significant difference between the percentage of standard tone pairs correctly perceived as ascending (M =.736, SD =.096) and the percentage of standard tone pairs correctly perceived as descending (M =.760, SD =.106), t(61) = 1.188, p =.239; d =.237. Tritone baseline ambiguity and priming. We examined whether or not tritones that participants perceived as ambiguous during the baseline trials were more susceptible to priming than tritones that participants perceived as unambiguous during the baseline trials. To do this, we identified tritones that each participant always responded to in the same way (always as ascending or always as descending) during the baseline. Similarly, we identified tritones that produced inconsistent responses (those that were sometimes identified as ascending and sometimes as descending) during the baseline. We then used a paired samples t-test to compare how often these two kinds of tritones were successfully primed. The mean percentage of tritones that participants considered ambiguous that were successfully primed (M =.485, SD =.059) was not significantly different from the mean percentage of tritones that participants considered unambiguous that were successfully primed (M =.484, SD =.065), t(57) = 0.117, p =.908, d =.016. Fatigue and practice effects. In order to determine whether participants were consistently accurate over the course of the study, we compared the accuracy of the standard tone pairs in the first block (M =.728, SD =.078) to the fourth block (M = 22

31 .761, SD =.083) using a paired samples t-test. Accuracy during the first block was significantly different from accuracy during the fourth block, t(61) = , p =.002; d =.410, such that participants were significantly more accurate during the fourth block. Discussion The purpose of this research was fourfold: to replicate the Simon effect and the tritone paradox, to demonstrate an illusory Simon effect with tritones, and to see if tritones can be primed. Results were mixed. We were able to replicate the tritone paradox but not the Simon effect. We also found partial support for an illusory Simon effect: response times to tritones were faster when the response was congruent with the ear of presentation. Finally, while some individual tritones may have been primed, there was little evidence of an overall trial-to-trial priming effect. Hypothesis 1 There was a significant interaction between hand and ear on accuracy, but not on reaction time, which indicates only tenuous support for this variation of the Simon effect (a decreased RT and increased accuracy for congruent trials). Further weakening this support was the specific pattern of the significant interaction on accuracy: congruent trials produced much better accuracy only when presented to the left ear; the effect for the right ear was considerably smaller (though still in the predicted direction). When considering the interaction between ear and hand, a combination of left hand and left ear in a trial produced slower response times and better accuracy; however, the combination of left hand and right ear in a trial produced faster reaction times and worse accuracy. These results may be the consequence of different speed-accuracy 23

32 trade offs on left- and right-ear trials. The high accuracy on congruent left-ear trials was accompanied by noticeably slower response times. Though this is the opposite of what a Simon effect would predict, it may simply indicate that quicker decisions lead to decreased accuracy. While this result did not confirm our hypothesis, it speaks to the nuances of the Simon task and suggests that the differences between the left and right ear in tone perception may be a good area for future research. There are several explanations for the above results. One potential explanation for the fact that we did not show a Simon effect in terms of reaction time could be because several participants only used one hand to respond to the tones rather than two hands as intended. In other words, some participants responded on the c key and on the m key with two different fingers on the same hand. We noticed that some participants were doing this only after the fact, so we were not able to correct them or to identify and eliminate all participants who may have done this from our sample. However, some previous research suggests that these participants may not need to be discounted because the critical left/right spatial relationship that produces the Simon effect is still present. That is, even though the responses were not produced separately by the left and right hands, they were located on the left and right relative to each other. Previous research has successfully demonstrated a Simon effect using non-standard responses that are not lateralized to different limbs, such as turning a wheel to the left or right (Wang, Proctor, & Pick, 2003). This supports the idea that we might still expect a Simon effect despite the non-standard motor responses. Although a Simon effect has still been shown in participants completing nonstandard versions of the Simon task, there is a chance that we would have seen a 24

33 stronger effect of congruency had all of the participants used their right index finger to press the m key and their left index finger press the c key. In future work, the instructions should explicitly state that one hand should be used to press one key and the other hand to press the other key, and researchers should monitor them to ensure that they do so. Furthermore, although most headphones and earbuds have an obvious distinction between right and left earbuds, some participants could have used older earbuds that didn t have right and left ear indications on them. If so, they would have heard all of the right ear trials in their left ear and vice versa. In future research, we would recommend including a safeguard (such as a trial requiring participants to respond based on the ear in which a tone pair is played) to confirm proper and uniform configuration of headphones. Hypothesis 3 For our third hypothesis concerning an illusory Simon effect, a congruent tritone trial was considered to be any trial in which the participant s response matched the ear of presentation (i.e., right hand and right ear or left hand and left ear). In contrast, an incongruent tritone trial was one in which the participant s response did not match the ear of presentation (i.e., right hand and left ear or left hand and right ear). Participants were significantly faster on congruent tritone trials than on incongruent tritone trials, which is consistent with the Simon effect. However, note in Figure 8 that the effect was not symmetrical. There was a large congruency effect for left-ear trials, but not for right-ear trials (and the latter was not in the hypothesized direction). The significant interaction between ear and hand displays evidence of a successful illusory Simon effect, consistent with the prediction of hypothesis 3. This 25

34 indicates that participants perceived tritone stimuli similarly to other stimuli that have displayed the Simon effect, in that participants responded more quickly to tritones they perceived as congruent than tritones they perceived as incongruent. Additionally, results showed that participants responded to standard tone pairs on average more slowly than they responded to tritones, which was an unexpected finding because one might expect participants to take additional time to understand ambiguous stimuli such as tritones. This suggests that, although tritones are inherently ambiguous stimuli, participants may not necessarily perceive them as such. Interestingly, although we were able to demonstrate a significant interaction between hand and ear for tritone stimuli, we were not able to demonstrate the same interaction for standard tone stimuli. This difference may have occurred because of how congruency was defined for standard tone pairs and tritones. Standard tone pairs were coded as congruent or incongruent based on the actual directionality of the tone pair, but no such directionality can be determined for tritones. Instead, tritones could only be coded as congruent or incongruent based on how participants responded to them. Perhaps an analysis in which standard tone pairs were coded by perceived directionality would show a significant interaction between hand and ear. This analysis would take accuracy rates into account. In most cases, perceived direction would be the same as the true direction; however, perceived direction and true direction would differ on trials in which participants made mistakes. Although hand had an effect on response time, the asymmetry was not based on handedness. Both left- and right-handed participants were slower when responding with their right hand. This difference was significant in right-handed participants, but 26

35 not left-handed participants. Therefore, handedness did not affect reaction time in the way that one would expect if use of one s dominant hand always led to a significantly faster reaction time. Hypothesis 2 For our hypothesis pertaining to demonstrating the tritone paradox, we also had mixed results. We were able to show that half of the tritones were perceived in different ways by different participants (and were therefore perceived inconsistently across participants). However, the other half of the tritones were viewed consistently across participants. This may not be surprising when one considers the homogeneity of our sample. Previous research has shown that tritone perception can be connected to home region (Deutsch 1991). Given that many of our participants were from near Seattle (27.9%) or Portland (21.3%), they may have shared a common linguistic palate. Thus, there is a chance that these participants could have skewed our data by all perceiving several tritones in the same manner (Deutsch, 1991). Our results showed that participants from the West Coast (California, Oregon, and Washington) as a group heard five individual tritones as consistently ascending or descending, whereas participants from everywhere else in the United States, taken together, heard only one tritone as consistently ascending or descending. Therefore, it seems that the (overrepresented) participants from the West Coast perceive a subset of tritones the same way, whereas participants from elsewhere did not have this same agreement. These participants presumably were a more heterogeneous group in terms of their accents and linguistic backgrounds. Consequently, some participants would have perceived a given 27

36 tritone as ascending while others perceived it as descending, yielding a larger number of ambiguous tone pairs for this group. Another potential explanation for the half of the tritones that were perceived consistently across participants could be unintentional priming during the tritone baseline. Though tritones are by definition neither ascending nor descending, most participants nevertheless perceive them as one or the other. Thus, it seems possible that tritones could have been primed by the perception of the preceding tritone. This would have made certain tritones appear less ambiguous than they actually were, especially considering that all participants saw the same sequence of tritones during this phase of the experiment. This problem may have been compounded by the lack of diversity in our sample such that if many participants heard the same tritone as unambiguous, they would have experienced the same priming effect. Such tritone priming during the baseline phase would also cause problems because these judgments form the basis for later comparisons. In order to avoid this possibility in future work, we recommend a mandatory delay between each tritone in the baseline (enough time to forget what the previous tone sounded like) and/or an auditory mask (such as white noise) played between each tritone. We also showed an unexpected effect pertaining to consistency in the perception of tritones. Our measure of accuracy for standard tone pairs required that participants consistently identify the tone pairs correctly (accuracy thus implies a level of consistency). While there is no correct response to tritones (and thus, no measure of accuracy), consistency for standard tone pairs and consistency for tritones are still comparable. We found a significant difference between the average scores for standard 28

37 tone pair consistency (or accuracy) and the average scores for tritone consistency such that responses to tritones were significantly more consistent than standard tone pairs. This could be due to the fact that there were only 36 tritones (12 tritones presented three times each), resulting in averages of 0.66 or 1. However, there were 240 standard tone pairs, so the average accuracy scores were much more varied and ranged below.66. It is surprising that people were more consistent with tritones than with standard tone pairs. Perhaps the averages would have been more similar had we included more tritone baseline trials. The recreated pitch class circles of three participants (Figure 9) showed an interesting pattern in that the pitch class circle of a participant from New Mexico was similar to that of a participant from Colorado, but very different from that of a participant from Washington. The pitch class circle of the participant from New Mexico had the D#-A interval as its vertical (or non-ambiguous) axis, which corresponds to our findings that the only tritone non-west Coast participants heard as unambiguous was D#-A. Note that tritones close to the vertical axis (highlighted in grey) are perceived more consistently (and, therefore, less ambiguously) from trial to trial than tritones closer to the horizontal axis. The D#-A interval was also close to the vertical axis in the pitch class circle of the participant from Colorado. By contrast, the vertical interval in the pitch class circle of the participant from Washington was F-B. This, and the other intervals close to the vertical axis (F-B, C-F#, and F#-C), correspond to the tritones that participants from the West Coast consistently perceived as ascending or descending. These pitch class circle differences further support the idea 29

38 that West Coast participants may have a different linguistic palate from non-west Coast participants. Finally, our graph of the percentage of tone pairs heard as ascending, shown as a function of the pitch class of the first tone in the ascending range, was very similar to the parallel graph from Deutsch (1987). This is exciting because we showed the same basic pattern using only three presentations of each tritone whereas Deutsch used 12 presentations of each tritone. Had we included more trials for each tritone, we may have found results that were even more similar to Deutsch s results. Hypothesis 4 Our results regarding priming were also mixed. In general, our results pertaining to ascending primes directly contradicted our hypothesis, as participants were more likely to perceive the tritone following an ascending tone pair as descending. This suggests a reversed priming effect in which participants perceived the tritone as different from the preceding tone pair. However, this same pattern was not obtained from descending primes. There was a trend toward perceiving tritones following descending primes as descending, but this trend was not significant. While there was no overall priming effect, a small number of specific tritones supported our priming hypothesis. Five out of 24 tritone-prime combinations showed the expected pattern in which participants perceived the tritone in the same direction as the prime. However, nine out of 24 tritone-prime combinations demonstrated the opposite pattern. These priming and reverse priming effects were more common among ascending-primed tritones than descending primed tritones. 30

39 It is possible that when a tritone was presented immediately after a clearly nonambiguous prime, the tritone may have appeared to be even more ambiguous by contrast. The participant may have been confused by the increased salience of the tritone s ambiguity and responded in keeping with a specific response set. In other words, participants may have chosen to respond in a specific direction when they could not easily disambiguate the tritone. It is possible that some participants in our study had a response set of responding to obviously ambiguous tritones as descending. This was reflected in further analyses, which revealed that five of our participants heard fewer than 10 of the 96 primed tritones as ascending. No participants showed a similar ascending tritone bias. There is a chance that these participants may have pulled down our averages. However, these participants were not discounted because they still had accuracy scores of above 60% for the standard tone pairs. Thus, we don t believe that they simply gave up or did not invest effort in the task. These reverse-priming findings could also be attributed to the selectiveadaptation technique, in which the alternate version of a reversible figure is more likely to be observed after adaptation to an unambiguous version of the figure (Long & Toppino, 2004). This effect takes place when the stimulus is presented for a long enough period of time that the neural structures supporting these versions become fatigued. In future work, we would recommend shortening the length of all of the stimuli (primes in particular) to examine the relationship between these reverse priming effects and the underlying neural structures. Finally, our results did not show a difference in priming between tritones that were perceived consistently and inconsistently. This may be a consequence of the lack 31

40 of a hypothesized overall priming effect. It also suggests that, if tritones can be primed, every tritone may be equally susceptible to influence from a preceding tone pair, regardless of whether or not a participant perceives the tritone consistently from trial to trial. In our analysis of priming, we considered the actual direction of the prime rather than the perceived direction of the prime, assuming that primes would always be perceived accurately. However, participants did make mistakes in their judgments of standard tone pairs. Consequently, misjudged tone pairs may have either diluted or counteracted any priming effect caused by correctly perceived primes. Thus, there is a chance that we would have shown a more robust priming effect had we used the perceived perception of the primes, rather than their actual direction. In future research, we would recommend analyzing the data in both ways in order to consider the differing effects of the actual and perceived direction of the preceding tone pairs. Previous research has shown mixed results pertaining to priming of tritones (Giangrand, Tuller, & Kelso, 2003; Repp & Thompson, 2010). Our results showed that, on average, perception of tritones could be manipulated, but only descendingprimed tritones were (non-significantly) primed in the hypothesized direction. Perception of ascending-primed tritones was able to be significantly manipulated, but participants perceived these tritones as descending. Although our results did not fully support our initial priming hypothesis, there was a significant change in participants perception of tritones following a standard tone pair. This supports a model of pitch perception that is influenced by external context and cues instead of an internal 32

41 understanding that cannot be swayed by context. This is similar to the way that individuals can perceive ambiguous visual stimuli differently from viewing to viewing. It is possible that we were not fully able to demonstrate a consistent priming effect because of the musical interval characterizing the prime. To be more specific, certain tone pairs were perceived as ambiguous (produced inconsistent responses) even though they were not tritones. Two intervals that illustrate this possibility are the perfect fourth and perfect fifth. The perfect fourth and perfect fifth are inverses of each other (i.e., moving clockwise five semitones around the pitch class circle from C to F provides an ascending perfect fourth while moving counterclockwise seven semitones around the pitch class circle from C to F provides a descending perfect fifth). Each differs from the tritone by only one semitone. Because these intervals span nearly the same distance in either direction around the pitch class circle, they are almost as difficult to judge as the tritone itself. Previous research has shown that the perfectfifth/perfect-fourth interval utilizes the same method of perception as tritones, in which proximity is not a useful cue for determining which pitch is higher, and in which there is a strong influence of pitch class on ascending versus descending judgments (Ragozzine, 2013). Because this interval is more ambiguous than intervals with more obvious distance differences between the two pathways around the pitch class circle (such as a major second of only two semitones compared to a minor seventh of ten semitones), it may not have created as strong of a priming effect on subsequent tritones. Additionally, it may have been the case that these more-ambiguous intervals resulted in longer response times, which would have decreased the strength of our prime. It has been previously shown that the effectiveness of primes depends on the 33

42 length of the prime as well as the amount of time between the prime and the subsequent stimulus (Wilton, 1985). Our experiment equated the number of times each tritone was preceded by an ascending or descending interval, but did not account for the size of the preceding interval itself. This may have created a confound that weakened our priming effect and should be a focus of future research. Strengths, Limitations, and Future Directions Our research had several strengths. For one, we used the same Shepard tones that Deutsch had used in her research (Deutsch, 1995). We also used the same methods as Deutsch in presenting the tritones (Deutsch, Kuyper, & Fisher, 1987). Thus, these results can be added to a body of data pertaining to the perception of tritones, using similar methods. Another strength of our study was that our standard tone pairs consisted of octave ambiguous Shepard tones, just like our tritone stimuli. This similarity led to perceptual constancy, reducing demand characteristics and making the ambiguity of the tritones less salient. Previous work that used unambiguous tritone pairs to prime ambiguous tritone pairs was unsuccessful in demonstrating a priming effect (Repp & Thompson, 2010). We also attempted to reduce demand characteristics by requesting demographic information, such as musical experience, after all of the trials. Nevertheless, there are several limitations that may have impacted our results. First, we encountered several technological issues. For example, in several instances, the tone pairs would not play due to software glitches, the internet browser, or the computer having been muted. This was especially prevalent during the baseline trials (the first tones heard in the experiment). In order to minimize the effects of this 34

43 problem, we discounted all tone pairs that produced reaction times longer than 20 seconds. Fifteen trials were removed from individuals tone pair averages for this reason (1 A-D#, 2 B-F, 1 C-F#, 8 D#-A, 2 F#-C, 1 G#-D). In future studies, we would recommend including some initial test trials to confirm that the computer s sound is functioning correctly and make any necessary adjustments before starting the tritone baseline. In addition, the experiment was lengthy, consisting of 390 trials. As with any lengthy procedure, there is a risk that performance changed over the course of the experiment. Our results comparing the accuracy of standard tone pairs in the first block and in the fourth block suggested a practice effect as participants were more accurate on later trials. This could be a limitation as we would expect participants to remain stable in the accuracy of their standard tone pair perception. The increase suggests that participants did not fatigue as the experiment continued. In fact, their performance may have actually benefitted from becoming more familiar and comfortable with the unusual Shepard tones, which most people have not heard. Nevertheless, standard tone pairs more heavily represented in the final block might have displayed artificially inflated accuracy. In future research, blocks should be counterbalanced to control for this effect. In future research, we would also recommend counterbalancing hand responses. It would be interesting to account for mental associations between the spatial orientation of the keys and ascending and descending. In our study, we assigned m (right) to ascending and c (left) to descending. However, in future research, 35

44 participants should be randomized into two groups (one group where m is ascending and one group where m is descending) to counterbalance response configurations. Final Thoughts There are several far-reaching implications of our results. For example, our results indicate a greater prevalence of absolute pitch in the general population than is typically assumed. Absolute pitch refers to the ability to name a pitch without the aid of a reference pitch (Deutsch, 1991). Although absolute pitch is generally considered rare, the existence of the tritone paradox indicates that most individuals do possess some form of this ability. The tritone paradox occurs because individuals hear certain pitch classes as higher or lower than other pitch classes. Although an individual may not be able to name the pitch class without the use of a reference pitch, there is clearly an underlying cognitive process that allows them to consistently identify and distinguish between different pitch classes. This process may differ from the ability to which absolute pitch traditionally refers because we were able to manipulate participants perceptions through the use of external cues. However, the basic ability to systematically distinguish between pitch classes without a reference can be thought of as a form of absolute pitch that is widespread in the general population. Second, these results draw a parallel between ambiguous auditory and ambiguous image-based stimuli, such that perception of both can be manipulated through priming. This can be demonstrated through the fact that a tritone pair was more likely to be perceived as descending than ascending when an ascending standard tone pair prime preceded it. This has important implications with regards to clarifying the nature of perception and manipulation of all modes of ambiguous stimuli. Future research should 36

45 continue to focus on the similarities and differences between these modes of perception. 37

46 References Bugelski, B. R., & Alampay, D. A. (1961). The role of frequency in developing perceptual sets. Canadian Journal of Psychology/Revue Canadienne de Psychologie, 15(4), doi: /h Deutsch, D. (1991). The tritone paradox: An influence of language on music perception. Music Perception: An Interdisciplinary Journal, 8(4), doi: / Deutsch, D. (1995). Musical Illusions and Paradoxes. San Diego CA: Philomel Records. Deutsch, D. (1997). The tritone paradox: A link between music and speech. Current Directions in Psychological Science, 6(6), doi: / ep Deutsch, D., Henthorn, T., & Dolson, M. (2004). Speech patterns heard early in life influence later perception of the tritone paradox. Music Perception: An Interdisciplinary Journal, 21(3), doi: /mp Deutsch, D., Kuyper, W. L., & Fisher, Y. (1987). The tritone paradox: Its presence and form of distribution in a general population. Music Perception: An Interdisciplinary Journal, 5(1), doi: / Deutsch, D., North, T., & Ray, L. (1990). The tritone paradox: Correlate with the listener s vocal range for speech. Music Perception: An Interdisciplinary Journal, 7(4), doi: / Fitts, P. M., & Seeger, C. M. (1953). S-R compatibility: spatial characteristics of stimulus and response codes. Journal of Experimental Psychology, 46(3),

47 210. doi: /h Giangrand, J., Tuller, B., & Kelso, J. A. S. (2003). Perceptual dynamics of circular pitch. Music Perception: An Interdisciplinary Journal, 20(3), doi: /mp Gopnik, A., & Rosati, A. (2001). Duck or rabbit? Reversing ambiguous figures and understanding ambiguous representations. Developmental Science, 4(2), 175. doi: / Long, G. M., & Toppino, T. C. (2004). Enduring interest in perceptual ambiguity: Alternating views of reversible figures. Psychological Bulletin, 130(5), doi: / Ragozzine, F. (2013). Correspondence in perception of the tritone paradox and perfect-fifth/perfect-fourth intervals. Music Perception, 30(4), doi: /mp Repp, B. H. & Thompson, J. M. (2010). Context sensitivity and invariance in perception of octave-ambiguous tones. Psychological Research, 74, doi: /s Scocchia, L., Valsecchi, M., Gegenfurtner, K. R., & Triesch, J. (2013). Visual working memory contents bias ambiguous structure from motion perception. PLoS ONE, 8(3), 1 8. doi: /journal.pone Shepard, R. N. (1964). Circularity in judgments of relative pitch. The Journal of the Acoustical Society of America, 36(12), doi: / Simon, J. R., & Rudell, A. P. (1967). Auditory S-R compatibility: The effect of an irrelevant cue on information processing. Journal of Applied Psychology, 51(3), 39

48 doi: /h Wang, D.-Y. D., Proctor, R. W., & Pick, D. F. (2003). The Simon effect with wheelrotation responses. Journal of Motor Behavior, 35(3), Wilton, R. N. (1985). The recency effect in the perception of ambiguous figures. Perception, 14(1), doi: /p

49 Table 1 Participant Demographic Information (N = 62) Demographic N Percentage* Gender Male Female Ethnicity White Biracial Asian Asian-American Hispanic Indian Handedness Right Left School Year First Year Sophomore Junior Senior Geographic Region of Childhood Pacific Northwest California Other within the US Multiple locations Born outside US

50 Music Experience with music Absolute pitch No experience with music Multilingual One language Two languages More than two languages *Note: percentages may not sum to 100 due to rounding 42

51 Table 2 Hypothesis 1: 2x2 ANOVA on Response Times to Standard Tone Pairs Source SS df MS F p 2 Ear *.069 Hand Ear*Hand Error Total *p <

52 Table 3 Hypothesis 1: 2x2 ANOVA on Accuracy of Standard Tone Pairs Source SS df MS F p 2 Ear *.179 Hand Ear*Hand <.001*.202 Error Total *p <

53 Table 4 Hypothesis 3: 2x2 ANOVA on Response Times for Tritones SS df MS F p 2 Ear Hand *.097 Ear*Hand *.105 Error Total *p <

54 Table 5 Hypothesis 2: Descriptive Statistics and Results of One-Sample t-tests of Tritone Perception during Baseline Tone Pair M SD Comparison Value 95% CI for Mean Difference t df p d A-D# , A#-E , B-F , C-F# , *.561 C#-G , *.369 D-G# , D#-A , *.966 E-A# , *.450 F-B , *.446 F#-C , *.432 G-C# , G#-D , *p <

55 Table 6 Hypothesis 2: Descriptive Statistics and Results of One-Sample t-tests Comparing Tritone Consistency and Standard Tone Accuracy Tone pair M SD Comparison Value 95% CI for Mean Difference t df p d A-D# , A#-E , *.417 B-F , C-F# , <.001*.769 C#-G , *.588 D-G# , D#-A , <.001*.677 E-A# , <.001*.867 F-B , <.001*.719 F#-C , G-C# , *.506 G#-D , *.545 *p <

56 Table 7 Hypothesis 4: Descriptive Statistics and Results of One-Sample t-tests Comparing Tritone Priming to Baseline Perception Tone Pair M SD A-D# (asc) Comparison Value % CI for Mean Difference -.143,.030 t df p d A-D# (des) , *.444 A#-E (asc) , *.576 A#-E (des) , B-F (asc) , B-F (des) , C-F# (asc) , <.001*.888 C-F# (des) , *.596 C#-G (asc) , C#-G (des) , *.446 D-G# (asc) , *.367 D-G# (des) , D#-A (asc) , *.537 D#-A (des) , *.490 E-A# (asc) , E-A# (des) , F-B (asc) , *.488 F-B (des) , *

57 F#-C (asc) , *.510 F#-C (des) , G-C# (asc) , <.001*.807 G-C# (des) , G#-D (asc) , *.436 G#-D (des) , *p <

58 Table 8 Hypothesis 2: Descriptive Statistics and Results of One-Sample t-tests Comparing Tritone Perception in the Baseline Between Participants from the West Coast and Rest of the Country (Comparison Value =.5) West Coast (n = 37) Non-West Coast (n = 18) Tone pair M t p d M t p d A-D# A#-E B-F C-F# * C#-G D-G# D#-A <.001* *.967 E-A# * F-B * F#-C * G-C# G#-D *p <

59 Figure 1. An example of an ambiguous image: some viewers see this image as a rat, others see it as a man (Bugelski & Alampay, 1961) 51

60 (Wolfe, 2014) Figure 2. Rising pitch helix, displaying the height and class of various pitches 52

61 Figure 3. Pitch class circle, adapted from Deutsch (1991) 53

62 Figure 4. The interval C-E is more commonly interpreted as ascending by four semitones (clockwise arrow) rather than descending by eight semitones (counterclockwise arrow). Pitch circle is adapted from Deutsch (1991). 54

63 Figure 5. Individuals can have differently oriented pitch class circles. In the first example, the individual hears B as slightly higher than F because it is closer to the highest point in the circle (in this case, A). In the second example, F is clearly higher than B. The third example is a tricky instance in which B and F are located at the same height within the pitch class circle and so the decision on which tone is higher is more ambiguous. Pitch circle is adapted from Deutsch (1991). 55

64 Reaction Time (seconds) Right Hand Left Hand Right Ear Left Ear Figure 6. Hypothesis 1: Mean reaction time to standard tone stimuli, measured with two factors (hand and ear) and with two levels each (right and left). Note that y- axis does not start at 0. Standard deviations are represented in the figure by the error bars attached to each column. 56