Music theory in the land of experimental psychology

The two readings I have assigned come from two different perspectives within the field of music cognition, the psychologist’s perspective (represented by Janata and colleagues) and the music theorist’s (represented by Martens). I consider both to be good examples of how experimental methods can enhance the study and theorizing of music.

At the same time, doing research at the boundary of two fields is not an easy task, and music cognition still faces many challenges, not the least of which is to find ways of bridging the methodological and conceptual gap between the different disciplines involved (psychology, cognitive science, and neuroscience on one hand, and musicology, music theory, ethnomusicology on the other) while cultivating questions that are sophisticated and relevant to all involved.

In reviewing these studies, focus first on the question that is identified by the researchers and how this question is operationalized into a testable hypothesis (i.e., how each element of the question is mapped onto some observable and measurable feature related to the phenomenon that is being investigated). Is the operationalization reasonable? What are the findings? How do the authors relate the findings back to the original question? What are some of the limitations of the experimental study that are identified by the authors?

Then, reflect on the question, findings, and limitations of the study from the perspective of your home discipline. Is the question relevant to your research? How might you use its findings? How might these two studies benefit from knowledge and know-how from your field of study? Conversely, how might your research (and your home discipline) benefit from the methods exemplified by these studies?

You might summarize one of the two studies, compare the two studies, or respond to one or more of the guiding questions. You might also imagine how you might collaborate with one of these researchers for a follow-up study. What would your study look like? What might be discussed in a “lab” meeting with the authors? Finally, you might also use your experience as a participant in the online experiments from my course to illustrate your points.

Post a response to the readings on the Forum blog by Sunday, November 24, 11:59 PM; post a response to one or more of your colleagues’ posts soon thereafter.

36 thoughts on “Music theory in the land of experimental psychology

  1. Imri Talgam
    Current trends
    Prof. Straus

    Note: The following response is by Imri Talgam. As a result of being unable to log on using my email address, I posted it using Javor Bracic’s login, with his kind permission.

    A Summary of Martens

    Martens’ experiment is an attempt to investigate whether the notion of tactus has any objective perceptual reality for listeners at large. As the author explains, the very choice of the term already involves some difficulties. Tactus is defined as the pulse that most people would choose to tap to in a given piece of music; this could be also often defined as the tempo, perceived tempo, or referent level. While these all refer to the same phenomenon, Martens denies that the tactus is an objective attribute of the score. Rather, he assumes that it is primarily a perceptual fact which varies from person to person, and attempts to construct an experiment to understand these variations.

    With this premise, he proceeds by trying to find an alternative model to resonance-curve theories of tactus. Common to these theories is the notion that listeners intuitively prefer pulses in a range of relative comfort (described by the resonance curve around ca 110 -220 BMP), and choose to decide on the tactus so the it falls within this range as a default. However, several examples of significant deviation from this show that there might be another mechanism governing the listener’s perception, which Martens sets out to find.

    Generally, the fastest consistent pulse (or FCP) in a piece is the most basic perceptual given that generates sensation of pulse. While listeners don’t always tap at this pace, preferring to group notes together as beats, it provides a good measure for finding out what tactus interpretations are possible, by defining ranges in which it is more and less salient, respectively. For example, an FCP of 400 BPM is unlikely to be considered a pulse, and is by default understood as a subdivision of a slower pulse.
    The alternative to the tempo-based resonance-curve model is then the subdivision (benefit) model, according to which a pulse can be established as the tactus if it is subdivided to a sufficient degree of consistency of 69% at the next level closer to the surface of the piece. This is in my opinion somewhat confusingly presented, since to my understanding “subdivision benefit” is not an act of subdividing a slower pulse in order to synchronize better (which is the common use of the term), but rather a the benefit of grouping notes together, so that the listener is not bound to the FCP. I thus read “subdivision benefit” as the benefit of understanding the FCP (or even a lower level) as a subdivision of a slower pulse.

    With this definition of the range of possible tactus choices (which is still constrained by the tempo ranges quoted above) and the conditions necessary to extract a pulse slower than the FCP, the experiment simply provides subjects with various musical excerpts while instructing them to tap to it.

    Here I’d like to pause for a moment to ask why this choice is never explained. How do we know that the pulse listeners tap is identical with their perception of tempo? Or, asked differently, what does is mean that they tap at a certain speed? Perhaps this sounds naïve, but it is really not intuitively clear to me, perhaps because as a performer I hear/use multiple layers of pulsation as a means of controlling larger rhythmical structures. Either way it seems to me that pulse means entirely different to someone who is a trained musician (and associates inflections of other musical parameters with the tactus choice) and a musically untrained person who taps without conscious definition of this act. Furthermore, any musician is used to thinking about a piece using different tactus choices a means of exploring it in a way which is far from intuitive. On a purely methodological level, bracketing these differences without explanation is somewhat baffling to me.

    The most significant result of the experiment is the definition of three distinct listener types, created by correlating each individual’s tapping choices with regards to the FCP and varying tempi. I will not describe this process in detail (although I find the idea of creating categories by clustering results interesting but questionable), but wish to comment on the interpretation offered.

    The three groups are basically defined by their relation to the FCP. The “surface tappers” mostly prefer to tap at the most convenient proximity to the FCP (or at the FCP when it is within the preferred tempo range). The “deep tappers” are the opposite, preferring to group attack points together to establish an underlying pulse far beneath the FCP, corresponding to what is usually thought of in music theory to hypermeter. Finally, the “variable tappers” change their relation to the FCP, presumably as a reaction to their more musically informed understanding.

    While this division can be useful as the basis for further research, there’s something I find rather disappointing about this set of conclusions. While it is suggested that surface tappers have a difficulty in extracting any deeper underlying metric structures, it is not asked why this is the case, as opposed to most trained musicians’ being either variable or deep tappers. At the start of the article, it seemed as if Martens was looking for a universal principle of perception of tactus; here, it is strongly implied that tactus choices are conditioned by musical training, and perhaps social differences. Before suggesting what social implications this might have, I would only say that this is again unnecessarily ambiguous.

    If we could assume for a moment that the ability to hear the FCP as subdivision testifies to the greater flexibility and consciousness of musicians, then the surface tappers can be considered to be less empowered, independent and critical towards their musical objects. This suggests that such distinctions can be used for a sociological model of different types of listeners, from the self-conscious professional to the naive consumer. This in turn could yield new questions for empirical research- for example, to what degree is the tempo-range of an individual determined by his/her consumption/involvement in a particular kind of music, what kinds of music tease the listeners to develop variable pulse choices and so forth.

    To summarize these reservations, I would say that the methodology of this highly interesting experiment basically suffers from a positivist view of perception, which brackets social/cultural conditions; as a result, the useful distinction between different classes of tappers doesn’t seem to lead to any new conclusions beyond the generality of the differences between musicians and non-musicians.

    • I’d like to clarify one point relating to your paragraph questioning the experimental method (tapping long a perceived pulse) and its implication for the perception of tempo (“Here I’d like to pause for a moment…”).

      Tempo perception is indeed a complex phenomenon, but I sense that part of the critique being made here stems from a misunderstanding of what it means to operationalize a question/hypothesis. There are many different ways to think about pulse, tactus, and more generally, someone’s perception of beat (if not metrical structure). When doing an experimental study, one measurable feature of this complex phenomenon needs to be chosen as representative. After analysis, findings can only be generalized with that specific operationalization in mind, and not with all possible ways of thinking/experiencing the phenomenon, i.e., the findings are “true” only for this particular form of tempo perception (with all its limitations). In essence, the researcher can generalize, but not essentialize, i.e., the phenomenon investigated cannot be reduced to the specific operationalization; I don’t believe Martens is making this mistake here (but I could be proven wrong: I’m always open to hear new evidence!).

      In empirical research, explanation of a complex phenomenon (which, basically, is what all musical phenomena are), will be reached through multiple (and I really mean many, many, many) operationalizations, not a single one. One single experiment cannot prove anything (actually, experiments cannot prove anything, period), nor can it settle a question by itself. It is only a baby step in one direction. If there is enough converging evidence, then a theory might come to be accepted by researchers as the best explanation so far for some aspects of a phenomenon.

      I’m also a little puzzled by the interpretation of this study as illustrating “positivism” on the basis of its bracketing off “social/cultural conditions” (last paragraph). I think I would like to hear more what is meant exactly by “positivist” here. But, I would caution the reader to avoid jumping to conclusions about the author’s assumptions. As mentioned above, one experimental study cannot be held responsible for answering all the questions one might have about the influence of social/cultural conditions (and to be fair, nor could a single music-theoretical/analytical study). The fact that Martens compares musicians and nonmusicians shows that he does not bracket off these conditions; the fact that he does not look into this in more detail here (i.e., explore other social/cultural factors) is most likely due to practical limitations of the population sampling and experimental design, not ideology. The experiment could be repeated with some modifications, looking at some other groups with the goal of exploring a set of social/cultural conditions (what is referred to as “between-subjects factors”).

      Also, even if it might appear obvious to readers that there would be different listening modes pertaining to pulse choice (as one possible operationalization of tempo perception), the identification of different groups of tappers in this study (i.e., tappers that most often choose one pulse over another in the specific repertoire represented by the stimuli) is a significant advance for psychologically-minded research on music perception. Before this, most studies did not take such a possibility into consideration, in part because they used non-musical stimuli (when the stimuli are more limited musically, they do lead to a narrower range of tapping responses). Martens’ study provides empirical evidence that listening mode matters a lot, and from there, other studies can be devised to explore specific aspects of these differences to gain a richer understanding of why something musicians “have knows all along” might be so.

      Ask me about hindsight bias during class…

    • Dear Imri,

      Thank-you for your comments. I think that I share some of your skepticism about the relationship between tactus and tempo, and you may be right that Martens did not adequately address this issue. I am especially sympathetic to your claim that musicians are “used to thinking about a piece using different tactus choices a means of exploring it in a way which is far from intuitive”; to me, part of the fun of music is trying out different possibilities for interpretation of its various parameters, especially its meter, and generally I think I don’t believe in there being “a meter” for a piece. When Ed Klorman visited last week to talk about theories of rhythm and meter in tonal music, he pointed out that one ought to spend a great deal of time thinking about meter, asking now just “how do I hear it” by “how can I hear it,” and trying out various possibilities, even the ones that seem counter-intuitive. Indeed, I’ve had stimulating discussions with friends over the issue of identifying the tactus of a piece of music, such as whether to hear something as a fast 3/4 with highly regular duple or quadruple hypermeter or as a compound duple (6/8 or 6/4) or quadruple (12/8 or 12/4). It’s interesting to try out the different ways of hearing things, and considering what factors might make it easier or harder to perceive a particular meter. Klorman also showed us how the perception of meter has changed over time, and may in fact be highly culture-specific: whereas today we tend to have a “German,” beginning-accented conception of metrical organization, there was once a widespread “Italian,” end-accented way of listening to meter that has been recently been rediscovered, and may have influenced the metrical theories of Hugo Riemann. But of course, musicians represent a very small segment of the population, and it’s worth investigating how listeners with no formal musical training perceive a tactus, whether there are any consistencies in their interpretations, and what these may tell us about the how the brain works.


      • I cannot resist but suggest a potential testable hypothesis based on your comment: “to me, part of the fun of music is trying out different possibilities for interpretation of its various parameters, especially its meter, and generally I think I don’t believe in there being “a meter” for a piece.”

        Experimental research has actually shown that you are right on target: people in general, and musicians in particular, are very good at imposing different “beat” structures on auditory stimuli (from undifferentiated signals to purely random event sequences). Not only are they able to do this, but when they do, it affects how they perceive the signal. For example, imposing a beat every 3 tones will make the first tone of the group sound louder and the gap between the last tone of the group and the first tone of the next group seem longer, even if there is no physical difference between any of these tones! Research has also shown that people vary in their ability not only to find a pulse, but the number of different pulse levels they are able to impose on a given musical surface, with more musical training generally resulting in preference for a slower pulse and access to more pulse levels. But that research has not been able to tell us that much about what features of the musical (note “musical” rather than just “auditory”) surface might interact with this ability. This is where more music researchers are needed!

        So, one potential design that could be explored and would potentially give rise to interesting findings in direction of your comment, would be to test how many different pulses are afforded by a given surface, followed by post hoc (after testing) analysis of the musical surface and theorizing of the observations. This would of course require many many such experiments to be conducted, but here’s the scoop: experimental research is usually a team business. Not need to do it all by yourself (and try to convince everybody else that your theory is better than theirs…).

  2. Megan Lavengood
    Prof. Joseph Straus
    Current Trends
    Nov. 24, 2013

    errata for Prof. Poudrier: the first experiment says “Please check this box is you can hear both sounds properly” during the sound test; I assume “is” should be replaced with “if.”

    As Prof. Poudrier suggested, I began with David Huron’s lecture and with the list review at the end of the article. I was surprised that he was engaging with postmodern critiques of empiricism; I actually would have thought that most people would brush such critiques aside as valid but irrelevant, and yet here is Huron engaging with it directly. I also thought this was a good refresher on important things to remember while looking at empirical evidence—while data is factual, interpreting the data is not, and it’s important not to fall into the trap of allowing your preconceptions skew the data. Huron’s other observations about what differentiates the humanities from the sciences were also interesting to me. He notes that humanities rarely take on false-positive skepticism (rejecting theories unless evidence in support of the theories arise) in favor of false-negative skepticism (accepting theories unless evidence of the contrary arises)—an attitude that one would initially assume to be a bad thing, since it is less rigorous and leads to unsubstantiated claims being propagated. But instead, Huron points out that this is a result of the field being data-poor; that is, we do not have enough evidence to prove much of anything, so being a false-positive skeptic in most humanities fields would get you nowhere. Huron encourages more understanding from the scientific point of view of these problems that the humanities have, and I am pleased that he does so. Huron also encourages the humanities to not react so negatively toward concepts of empiricism and truth content, which again is something that deserves to be said. Essentially I agree with everything Huron is writing here and I appreciated his eloquent contrast between humanities and the sciences.

    To summarize the article by Janata et al., there are two hypotheses being tested. The first study tests the hypothesis that things like tempo, genre, familiarity, and enjoyment affects the “groove rating” of a given piece. I suppose this makes it a correlational study, according to Huron’s types of empirical studies. The results stated that genre and tempo had a large effect on groove ratings, showing unsurprisingly that songs in the R&B/soul genre and songs that have a fast tempo have the highest groove ratings. Enjoyment also correlated with groove ratings, but familiarity did not. Limitations of this study are probably the need to quantify things that are not typically quantified: we don’t typically think of a song’s “groove rating”; we just groove to it! Explaining it with a dial or a number is rather artificial, but necessary if one is to do a measurement study. The second study tested whether sensorimotor behaviors increased when the groove rating was increased. This hypothesis was confirmed, as participants felt as much in the groove while tapping to silence as while tapping to low groove examples, and participants also engaged in far more sensorimotor behaviors when high-groove examples were being played. These experiments, along with other surveys conducted, confirmed that groove has sensorimotor coupling with the music at the heart of its definition. Another interesting finding was that being allowed to freely tap to the beat did not increase the feeling of being “in the groove” as the authors had expected it would. The authors also found that songs having a groove increases the sense of enjoyment one feels from listening to the piece.

    Martens’s article can be summarized as testing the hypothesis that the tactus in music becomes clarified in the presence of a subdividing pulse; the listener will choose a tactus that has at least one perceptible level of subdivision, and not the fastest consistent pulse. Martens asked subjects to tap along to musical excerpts. He tests this by playing various stimuli and noting with which pulse level various listeners tap along, and comparing these results among all the listeners. He verified that while tactus is certainly constrained by tempo and listeners mostly do not tap with the fastest consistent pulse, it actually is not clarified by subdividing pulses; on page 440, Martens writes, “Of the 781 successful tactus choices, only 439 (56%) were a subdivided pulse.” Martens groups listeners into deep listeners (who choose a level much deeper than the fastest common pulse), surface listeners (who often choose the fastest common pulse), and variable listeners who switch between the two, but he is careful to note that people may change their listening patterns based on environmental factors. The main limitation of this study, which was acknowledged by Martens in his conclusion, is that it is impossible to determine to which features a listener attends when determining tactus without a more controlled study, but that this study would then be “less ecological,” which I presume means that the audio stimuli would not be real music but instead some kind of computer-generated audio.

    My own research deals somewhat with rhythm and meter, as I am interested in the music of the indie pop artist Sufjan Stevens, who often chooses very lively tempos, asymmetrical meters, and relatively complex rhythms. Although these hypotheses tested by these authors does not directly relate to my research, they do lead me to interesting questions. One feature I find remarkable about Sufjan’s music is that while he uses asymmetrical meters, I rarely notice this feature unless I am specifically listening for it; in other words, his use of asymmetrical meters sounds quite fluent and not at all forced, taking away the lopsided sound that would call attention to its asymmetry. But I’ve found this perception to not be universal—other people notice the asymmetrical meters right away. Perhaps such a difference could be connected to a difference in perception of tactus; perhaps I am attenuating more to the fastest common pulse (e.g., the eighth note in a measure of 5/8) while the other listener is attending to a pulse that is not even (e.g., the dotted-quarter followed by the quarter in a measure of 5/8). I think the Janata article could be expanded in a number of directions, but one direction that is most relevant to my research on pop music and focusing on Sufjan in particular would be to study groove ratings of pieces with asymmetrical meters, either on their own or in comparison to pieces with more standard meters.

    • I’m glad you found Huron’s article useful. I think that he makes some excellent points and since I read it a while ago I have been doing a lot of thinking about the implications of the “moral dilemma” of researchers, and how to better address it in my work and in our field.

    • Hi Megan,

      I also found Huron’s article to be eloquent and rather intriguing. I agree with you that he does a great job summarizing a few differences between the humanities and the sciences. An aspect that I found rather remarkable was his point that saying ‘I don’t know’ or ‘We cannot know’ is no more epistemologically safe than “the corresponding positive claims ‘I know’ or ‘In principal, we can know.'” I think I shy away from empiricism because of the tension (which I have written about in previous responses) caused by our own subjective views of the art we analyze.

      Thanks for your comments about the use of the term “groove.” For some reason, I was approaching the concept completely from the perspective of a composer and trying to imagine what I might do to create a groove. This is what caused me to think of it as a music-theoritical concept, but I completely agree with your description! When I consider how most of my friends use the word, it really has little to do with a strict theoretical idea!

    • Dear Megan,

      I decided to write a response to your thoughts on Poudrier’s readings since I haven’t yet heard from Christina, one of my study group partners. Also, I was never paired with you for responses during the course, so I am taking the opportunity to respond to someone that has not been in one of my groups.

      First and foremost, I thought Huron’s lecture was wonderful—excellent musings on empiricism and the distinct forms of skepticism typically associated with the humanities versus the sciences. I, like you, found the material to be incredibly thoughtful and well-written.

      On the Janata study, I tend to agree with you that the use of a slider, or number, to indicate “groove rating” seems slightly superficial, but I wonder how Janata and crew could have arrived at relative groove ratings in another way; as you note, they need to establish relative “grooviness” in the first part of the study in order to proceed to the meat of the matter: the correlation between perceived groove and the compulsion towards bodily movement. I also wished I had had the time to listen to all of the excerpts that they included (I have to admit, the inclusion of Stevie Wonder’s “Superstition” as a quintessentially “groovy” track did get a chuckle out of me while reading). Conversely, with the Martens study, I was more convinced by the tapping exercise as a concrete and measurable phenomenon for analysis, but was concerned that the focus on that one body part might be missing part of the bigger picture, in terms of the bodily expression of tactus choice. In thinking about the two, I even began to wonder if different parts of the body might tend to “count” in different ways (i.e. might the legs and feet count deeper in comparison to the arms, hands, and fingers).

      In citing limitations of the Martens study, I noted that it was incapable of discussing musical stimuli that involve significant changes in tempo, meter changes, uncommon or asymmetrical meters, and metric modulation—all things that I think would be interesting to study, but which fell outside of the scope of this particular study from the get-go, based on its methodology. I myself seem to be very sensitive to asymmetrical meters, and especially in listening to popular music (I suppose, in the context of pop, asymmetrical organization seems even more surprising to me than in the context of “concert music”), so I found your comment about Sufjan’s music interesting. I’m not incredibly familiar with his music, but I do very much appreciate sophisticated pop music, so you have definitely given me a reason to start investigating him.

  3. Tom Johnson
    Prof. Joseph Straus
    Current Trends
    Nov. 24, 2013

    For my response this week, I will provide a curt summary and critique of Martens followed by some thoughts on the experiments we took part in this week. This was my (and I’m sure many other students’) first relatively in-depth encounter with the cognition side of music theory, and I found it sort of challenging to engage with. This was not so much because of the technical and statistical means of analysis, but more because of the methods, questions, and aims of the experiments. It is a difficult thing to dissect one’s own understanding of how one engages with music, especially in such basic ways as finding a pulse; it just sort of happens! It is probably for this reason that I found myself wondering why the researchers chose to focus on various minutia or why they investigated correlations between some aspects of their experiments (eg. Martens’s spontaneous tempo and mean tapping rates). In any case, I found it an interesting week of reading of a style very different than some of our other more traditional music theory topics! (As an apologia, I would like the reader to know that I’ve been sick and in bed the past couple of days, but I hope the coherence and intellectual level of my responses aren’t too affected by my feverish state!)

    Martens Response
    Basically, Martens sets out to test how a listener’s response to meter and tactus is affected by the subdivision of pulses and by the tempo of music. In many theories of rhythm and meter (specifically Hasty and London), subdivisions are understood to refine one’s sense of meter and beat, helping the listener to home in on a more accurate prediction of the next beat by providing more rhythmic data. I was a bit confused by the initial discussion of how the percentage of articulated subdivisions correlated to tactus choice, and it seemed as though the 69% threshold was a bit…arbitrary. Though I lacked the time necessary to read the Patel et al. (2005) paper, Martens notes that the stimuli used to measure this threshold were “not fully musical,” so I’m unsure as to how exactly this limit was reached. In his Figure 2 (p. 434), Martens gives various rhythms that contain different saturations of subdivisions. For me, the eighth notes in each case do indeed help to refine my sense of tactus and beat, regardless of the exact percentage of articulated subdivisions. Perhaps I missed something here, but I felt this portion of his background section suffered a bit from a lack of coherence and could have been fleshed out a bit.

    The comparison Martens made between a spontaneous tempo and the mean tapping rate of his subjects seemed a bit, well, silly to me. I understand that he wanted to separate out any correlations between chosen tactus and subject background from the relation of chosen tactus to the musical stimuli, but this measure seemed a bit contrived. Indeed, he found that these spontaneous tempi “did not seem to direct those subjects’ choice of tactus in the musical excerpts.” (p. 438) This seemed like common sense and an unnecessary step to me, though I of course understand that part of the point of cognition studies is to show how our common notions of musical perception actually play out empirically, and results that differ from our expectations are often the most consequential theoretically.

    It was rather clear to me that the surface tappers (those that consistently followed the FCP) didn’t really understand the instructions clearly, as Martens addresses in his footnote 4: “I am certain that with minimal instruction, most, if not all, Surface Tappers could be led to hear past the surface pulses.” (p. 446) (A similar thing happened to me in the third of the online experiments for this week [“Tapping to Uneven Beats”] I believe. I thought we were supposed to tap along with one of two simultaneous, uneven rhythms, but I think the point was to tap a consistent, larger metric pattern; the tactus in Martens’s study really.) As Martens mentions, the surface tappers had the least amount of musical training of the subjects, so additional instructions likely would have helped them understand what the actual aim of the experiment was. Conversely, I wonder if it would have been beneficial to have the subjects tap their feet rather than hands; it would be harder to keep up with the FCP in higher tempi, forcing the subjects to keep time in larger groups. Or perhaps, in a more imaginative approach, subjects could dance in whatever way they feel is appropriate to the music at hand; then other subjects would interpret the dancing to determine which beats the dancing subjects emphasized. Regardless, if I were Martens, I would have addressed the issue of surface tappers briefly at the outset of the results sections and then focused my data analysis on the other two groups, basically relegating the surface tappers to a brief paragraph.

    Along these same lines, I was glad to see a discussion of the “variable tappers” alone in Martens’s results section. These were the subjects whose choices in tactus were divided between subdivided pulses and FCPs depending on musical context. By comparing to the consistent deep and surface tappers, Martens was able to shed light on some of the musical features that helped guide the variable tappers towards a higher or lower level pulse.

    Overall, Martens seemed to confirm what many of us in music theory might suspect: different listeners with different musical experience will consider different levels of pulse to be most salient. But the consistency with which listeners chose either the FCP or the SP as tactus, regardless of tempo or musical context, was surprising to me. As discussed with the surface tappers, I think this could be partially explained by the clarity of instructions, which were intentionally somewhat vague in order to gauge the participants’ unbiased responses. I think a higher level of specificity in the directions could have led to more intriguing and illuminating results in terms of tactus choice.

    Brief Discussion of “Perceived Emotion” Experiment
    As an aside, I would like to comment briefly on one of the experiments we took part in this week. In “perceived emotion of musical excerpts,” the subject was asked to rate the activity and the relative quality (positive or negative) of the music in addition to selecting various adjectives (eg. happy, sad, calm, excited, etc.) that described the excerpt. The excerpts were played about three or four times each with different tempi each time. I took issue with the methodology here. When the listener hears the same excerpt a second time, he/she immediately compares it to the prior hearing. This affects how the excerpt will be rated the second and third times, and I believe creates an unwanted amount of self-reflection that I think might bias the results. (“Oh, I heard this one before, but now it seems to be slower…how did I rate it before? The experimenters must be trying to determine how tempo affects emotion…Since I know this, will I change my answers? Surely the activity is lower since the tempo is slower…or is it? This was happy before…but is it joyous not? What is the difference between calm and content, or happy and joyous?”) I’m curious as to how the successive ratings will be analyzed after they’ve been collected. I personally would have played each excerpt once for each listener, with different tempi for each. This way, there could be a basic control group whose subjects heard each excerpt at the normal, recorded tempo. The results of the others would be compared to that control group, rather than compared to the “control” of the original hearing in the current study. I’m not sure if I worded this clearly, but I would like to hear it addressed during out class meeting.

    • Unfortunately, I am not the best person to lead a discussion of this online experiment because it was designed by one of the two student groups. I will alert them of these questions and maybe they can provide some responses. One important thing to note: These group experiments are the very first experience most of my students had to build an experimental study from scratch. Also, this is exactly why it is always a good idea to do a pilot study before you run the “real” one!

    • Thanks for your thoughtful response, Tom. (I’m one of the students who designed the experiment you discuss above.) In short, I agree with you 100%: the methodology is problematic. When we were designing the experiment, we discussed the exact issue you point out. Our decision to play each subject all of the excerpts was made, more or less, out of necessity. Since we were anticipating 10–12 participants for our pilot study, we decided to maximize the amount of data at the expense of methodological rigor. If we were to run this study “for real,” we would proceed with a design similar to the one you suggested (i.e., with a proper control group).
      All of that said, we did run a pilot experiment last week in which the examples were played only at the original tempos. And so we have something akin to a control group, though it’s not air-tight (for starters, we had only 5 participants!). The data from that study confirmed (again, with only 5 participants) that the examples we chose to represent specific valence-arousal spaces did correspond with the expected emotions.

    • A couple additional comments:

      You state: “It was rather clear to me that the surface tappers (those that consistently followed the FCP) didn’t really understand the instructions clearly.” What is the basis of your certainty on this point? I can imagine several other possibilities, including the fact that some listeners might just generally prefer tapping faster than others (again, just one possibility). Of course, one could come up with some design to test an hypothesis generated by your conjecture, but another alternative that is less “costly” is to debrief your participants (either in person or with a questionnaire).

      As for the online experiment, the team’s goal was indeed for participants to tap to either one of the two uneven meters or some other combination of pulses, and not only to a larger meter.

      I think your suggestion of using the feet as another tapping condition is a really good one. There might indeed be some motor system factors at play, and this would be a very productive way to tackle this issue.

      • I suppose my certainty was at least partially affected by my cold over the weekend! Really, the comment was partially in response to my own failure to understand the directions in the third of our experiments this weekend. I ended up tapping along with one of the stimuli rather than imposing my own metrically-derived tactus. I think my problem with the methodology in Martens really lies in the interaction between the instructions, the synchronization task, and the surface tappers. I can’t seem to find where Martens explains the synchronization task, but it certainly seems plausible that it could have elicited a tactus of FCP from the surface tappers, which they then interpreted to be the salient pulse level for the rest of the experiment. Perhaps the FCP was relatively “comfortable” to tap along with in that excerpt, and surface tappers, without much musical background, assumed that the FCP should be what is comfortable.

        Thank you for your comment. I agree, I am definitely not in a place to support my certainty in the surface tappers’ lack of understanding; but I do think I am somewhat justified in questioning the relationship the instructions (esp. with the term “comfortable”), the synchronization, and the surface tappers. Perhaps another tactus experiment with slightly more suggestive instructions or with a number of different synchronization examples would provide a useful comparison to Martens’s results.

        • Dear Tom,
          I am glad you mention the difference between the music-theoretical approach and the cognitive approach to discovering truths about music. It was also quite hard for me to try not to utilize my speculative instincts, but stick to the bare facts, data and only the most careful conclusions possible. This approach is very valuable, though, as it can dispell myths and legends about what good music is and what good music theories are. Again, we sometimes might be connecting more dots than there really are…
          The idea of dancing to the music seems great to me – there should be more studies about dance – which probably is the most natural bodily response to music! But it seems to me that it might be quite hard to interpret the findings, as different people will judge the moves of the dancer differently, and that would only increase the probability of making false conclusions.
          I was also happy to read about the variable tappers in Martens’ study – I believe that this group is the most open-minded and musical group. They were ready to change their convictions, and react to the textural patterns of the music to lead them in finding a different tactus each time, without relying on a preexisting schema. I could see them as being imaginative musicians and conductors.
          I also see your point about giving more specific instructions to tappers to make sure they know what is expected of them – I generally agree, but I also see how that could shift the results all in one direction, and render the whole study useless.

    • Dear Tom,

      Thanks for your thoughts on the Martens article. I also have only limited experience with the kind of research with which we were presented this week, so it’s good to see what other people thought about the same article. Your point about the arbitrariness of the 69% subdivision threshold is well taken. I wonder if that threshold varies from listener to listener, what factors contribute to a higher or lower threshold, and whether an experiment could be designed to investigate such questions. I hypothesize that the threshold could be reduced through musical training, since the ability to project a steady beat comes with practice, but I imagine that there are many other variables at play.

      I skipped over Martens’s discussion of the spontaneous tapping in my summary of his article because he seemed to waver on whether or not there was any connection with the activity at hand, so I am thankful that you addressed it. It seemed a bit overkill to me, too, but I think I might know where he was going with it. Some people are really wired all the time (too much caffeine or whatever), so maybe that would come out when they go to tap something spontaneously. If that person then listens to something and has to tap along, maybe it’s the case that they’ll have to find some kind of balance between their baseline rhythm and any of the reasonable choices of tactus. Same goes for people who are chill as ice. I think it would be an interesting extension of the experiment to see what happens if people are asked to perform the same task after 15 minutes of cardio or a double espresso on the one hand, or a Thanksgiving Day dinner on the other.

      Something you said about providing more instruction to the surface makes me wonder about something else. Maybe some of them were keeping the tactus with their foot while their hands were drumming the surface rhythm? Did anyone think to check what the subjects were doing with the rest of their bodies?

      Lastly, thanks also for sharing your experience with this week’s online tests. I also found the perceived emotion test very frustrating for some of the same reasons. Additionally, I did not know what to make of the spectrums of active/inactive and negative/positive. For example, one of the excerpts we listened to was something French-sounding. It started out with a languid minor interval, but went on something major toward the end. Is that negative or positive? It started out one way, maybe but then turned went another… without the context, I can’t make heads or tails of it. Another example had a bunch of running notes, but the harmony stayed the same. Is that active or inactive? Well, it’s both. Do they cancel each other out? Don’t even get me started on the “emotions”! I won’t belabour the point, since it seems that the misgivings have already been addressed. I’ve no doubt we’ll all be talking about our experiences tomorrow, anyway.

      In the meantime, I hope that you’re feeling much better!


  4. Response to Musical Cognition
    By Alexander Martin

    For this week, I will be responding to the Martens article on ambiguous tactus and listening strategies. The study’s starting point is that when people tap along with a piece of music, what they are tapping corresponds to the listener’s main choice of beat—this is the tactus. What metric level is chosen as tactus will vary from listener to listener. Although it had been assumed that there was a goldilocks zone around 100 to 120 bpm in which most people will choose a tactus, studies have shown instances where a preponderance of people will favour a tactus outside of that zone for a given piece. Clearly tempo is not the only factor, so is there a better predictor for what people will choose as tactus?
    Martens originally wanted to develop a model that ranked our choices of metric structure via the concept of subdivision benefit, which has been surmised by other authors. Subdivision benefit is the idea that if a pulse layer is consistently subdivided, then its salience in the music is reinforced. This assumes a hierarchy of attacks that is familiar to one of Lerdahl and Jackendoff’s rules of well-formedness: beats at slower layers must also be beats at the next faster level.
    Since it is often the case that the surface level of the music is inconsistently subdivided, Martens is interested in knowing what degree of consistency is required in order to hear a given level as a beat. Prior studies suggest that we’re willing to accept quite a bit of irregularity before rejecting a layer as inconsistent: according to Montague, “the implication of regularity is more important than its actual occurrence.” Thus, Martens introduces the concept of a consistency threshold, and sets its value to 69%: “periodicities in which less than 69% of beat-points coincide with are unlikely to be felt as beats […] and are unlikely to contribute to a slower pulse’s salience via a subdivision benefit.” Any given rhythmic stream needs to exceed this level of consistency before it can be considered subdivided.
    The consistency is further constrained by the tempo. According to Parncutt, our sensation of pulse falls within a 33–300 bpm window. Martens conservatively adjusts the upper boundary of this window to 240 bpm in formulating his hypothesis (see next paragraph), citing a drop-off in our ability to feel pulses at higher tempi. This means that any pulse slower than 66 bpm will not be able to be felt as a subdivision of a higher metric level. What remains are the fastest consistent pulses or FCPs between 66 and 240 bpm.
    The purpose of his experiment is to determine whether people’s choice of tactus could better be predicted on the basis of subdivision benefit or tempo. Martens hypothesised that: “in music that has a least two consistent pulses faster than 33 bpm, a listener’s tactus will usually be a consistently articulated pulse that is divided by at least one other consistent pulse,” i.e. not the FCP.
    So how was this hypothesis tested? In the actual experiment, subjects listened to a recorded musical example and were asked to tap along with it. Subjects had diverse backgrounds and levels of musical training. The examples were all about 20 seconds long, drawn exclusively from the Western classical canon, and covered a range of tempi. Eleven out of thirty examples had an FCP above 240 or below 33 bpm.
    For most of the pieces, listeners chose between two possible choices of tactus (corresponding to different levels of metric structure), but a handful of pieces elicited three and four different choices of tactus. Almost all of the tapping occurred within the 33–300 bpm range, as expected. However, a surprising number of tappers remained loyal to the FCP, even when there was a pulse stream at a lower tempo that would have benefited from subdivision. The author concludes that the choice of tactus was affected both by the excerpt’s tempo and metrical structure.
    But what about the supposed subdivision benefit? The author concludes that it exists for some but not all listeners, and not equally, besides. The data collected shows that some listeners were happy to choose the FCP as tactus even when the tempo was very fast, while others were happy to pick a subdivided pulse (SP) as tactus. Based on these findings, Martens proposes that there are basically three different listening strategies that might guide someone in making a choice about tactus. The listeners who prefer to tap the surface rhythms, he calls surface tappers, while the tappers that pick out a SP (and for whom subdivision benefit correlates positively), he calls deep tappers. Finally, the tappers that pick the FCP as tactus about as often as chance would dictate, he calls variable tappers.
    Martens observes a correlation between a subject’s listening strategy and his or her reported level of musical training: while 40% of the surface tappers reported no musical training, all of the variable and deep tappers reported a minimum of six years’ worth of training. Even more interestingly, the deep tappers reported training in areas that would seem to require a greater awareness of higher level metrical structures (for example, conducting, low-range wind and strings, and composition), while variable tappers had experience in areas that arguably require an awareness of both background and foreground rhythmic structures (for example, piano, high winds and strings, and guitar).
    Martens’s hypothesis that a familiarity with faster moving surface textures has an impact on one’s choice of tactus resonates well with my experience as a clarinetist and amateur guitarist. I frequently tap along with music when I listen to it. I can’t remember exactly when it started, but at some point I began to become more self-aware about my choice of metric level. Although I haven’t done any empirical research toward the matter, introspection tells me that I usually benefit from a tactus that can be subdivided. Partly the reason for this is that I like to improvise rhythms (often to melodies in my mind’s ear) to go along with the music I listen to, and it is easiest to improvise at the surface level while retaining a higher pulse as tactus. Along the same lines, sometimes I like to change artificially my choice of tactus to see what impact it will have on my experience of the music. This is especially enjoyable at moments of arrival or change in music with which I am already familiar. My experience of mixing it up with the tactus seems to validate Martens’s later point that modes of listening are not fixed. To choose the road less traveled from time to time is a breath of fresh air; Martens’s research does a good job mapping the likeliest alternate routes.
    As to research implications, the one that comes readily to mind is in determining what Caplin sometimes calls the “real measure.” It is sometimes difficult to determine whether the idea or phrase level lines up with the measure, or the two-measure, or the half-measure metrical level. In clear cut cases, Caplin notates that the written meter is not representative of the perceived tactus by writing, for example, N = 2R (notated measure = 2 “real” measures). However, things are not always so clear cut, and I can recall several instances in a class on form last year where the matter was debated in class (and on midterms). Martens’s studies provide some new terminology with which to address such problems. An ambiguous passage might be clarified by considering the consistency threshold. For pieces with several layers of consistent sub-divisible pulse, different analyses could be made with different listeners in mind.

    • Tom Johnson
      Response to Alex

      Dear Alex,

      Thanks for your very lucid summary of the Martens article. After reading your response (and after enough rest to get over my cold!), I definitely understand his methodology much more clearly. I was intrigued to read about your conscious tactus and metric games in your improvisations. What a music-theorist thing to do! Personally, when listening to a straightforward beat, I often try to tap out somewhat irregular time signatures (usually a 5-beat, though occasionally a 7-beat meter) to create tension and to “investigate” the interaction between it and the more typical meter. This means, obviously, that I am engaged in a deep tapper strategy when I do this, since I rely on a whole host of subdivisions to attempt to line back up at a higher metric level. Of course, this is only possible with music or styles of music that are relatively familiar to me, and is enjoyable to me when I am able to make it work out. This, I believe, lines up with your experience concerning the moments of arrival or change in familiar music when you artificially change up your tactus in improvisation. Indeed, the road less traveled can be a refreshing one to take, but it requires a road that is more traveled!

      I am not familiar with Caplin’s ideas of “real measure” or perceived tactus, but I can see how problematic this might be. Recalling Ed Klorman’s visit last week, even familiar music can be metrically confusing at higher levels; Martens’s article, though only concerned with the beat level instead of various, shifting hypermetrical levels, confirms just that. I wonder how meter and tactus might be studied from a sociological perspective…

    • Thanks for the concise summary of Martens’ experimental goal (i.e., “The purpose of his experiment is to determine whether people’s choice of tactus could better be predicted on the basis of subdivision benefit or tempo.”) I think we can all agree that within the confines of the operationalization, the answer to Martens’s question is “no” (or at least, it is that FCP is not the best predictor of tapping behavior).

      So what? Well, your commentary suggests possible avenues for further research, and there are many more. It is often the case that one experimental result will generate many more questions. Possibly against popular belief, empirical research might not be for those who are looking for unshakable evidence that this or that is the truth. Rather, it turns out to be an ideal playground for insatiable curiosity!

      Closer to home: A great exercise upon reading a study like Martens’s is to make a list of questions generated by the findings. And given that music theory is such a friendly field, one might even send some of these to the author! A fruitful collaboration might ensue…

      • Dear Alex,
        Thank you for a great summary of Martens’ Study. I like how you relate it back to our readings by Lerdahl and Jackendoff. Also, I was pondering the Montague quote that you mention – it seems as though we are wired to see order and regularity all around us – there must have been an evolutionary benefit in connecting the dots even where they do not connect. It seems to be a profound and central human capability to make order out of chaos.
        The result that most of the variable tappers and deep tappers were trained musicians was intriguing. Does that mean that we are somehow taught to listen for the SP, but that it is not the natural way of listening to music? You bring out a good point in saying that tapping the larger rhythm is beneficial for improvising in-between the taps. It also helps to fathom the work as a whole much easier. But it surely crumbles as soon as the phrase rhythm changes. This is where a tactus closer to the surface might be more unified and more beneficial. I was reminded of someone who used to play Bach LP records at a faster speed, and even in spite of the distortion of the pitch, it was somehow easier to comprehend the long-term sections and harmonic developments. Excellent (if sacrilegious) experiment!
        In the end, the conclusion of all of our readings this semester seems to hone in to one central idea that there are no generalizations, and we shouln’t attempt to make any. Humans perceive music very differently. But figuring out the details of these differences can be exciting!

  5. Simon Prosser’s Response to the Readings

    I don’t know a great deal about music cognition (or the cognitive sciences is general), or the psychology of music, and in general, I feel that my own research interests do not intersect much with research on the cognition and psychology of music. But I am curious, and so I am glad to have had the chance to be exposed to some of this research so that I can learn more about its methods, its goals, its findings, and what it can tell us about music and the way we perceive and interact with it.

    Peter A. Martens’s paper on “the ambiguous tactus” investigated the effect of tempo and beat subdivision on a listener’s ability to identify a tactus in a passage of music. This was tested by having participants listen to a series of short music excerpts from a wide variety of styles and tempi while tapping along with their hand on a sensor that would record their taps. The participant’s tapping would represent where they perceived the tactus. What was found was that the perceived tactus in a particular passage of music varied greatly from one listener to the next, with no clear correlation with tempo or with subdivision; however, there did emerge evidence of three “listener strategies”: “(1) tapping with a subdivided pulse, (2) tapping with the fastest consistent pulse in the music (a pulse with no consistent subdivision), or (3) using a mixture of these two strategies based on inconsistent rhythmic activity at the musical surface.” (433)

    In Petr Janata, Stefan T. Tomic, and Jason M. Haberman’s paper, the authors investigate the concept of “groove” as a psychological construct, something that individual listeners understand intuitively as a physical “coupling” with the music. This was accomplished by having participants complete a survey in which they were asked to describe what they understood by the term “groove,” and then listen to a series of short musical excerpts drawn from several different contemporary genres and representing a variety of different tempi; sometimes the participants were instructed to make no physical movements while they listened, sometimes they were asked to tap along with their hands, and sometimes they were given the opportunity to “free form” their physical response to the music. “The construct that emerged, both from open-ended definitions provided by the participants and from the endorsement of a large set of statements, has at its core an emphasis on sensorimotor coupling with the music and positive affect,” and “we showed that perceived groove is related to a sense of enjoyment.” (70)

    Prior to this week, I would say that most of my exposure to music cognition and psychology had come from Lawrence Zbikowski’s book, Conceptualizing Music. Admittedly, Zbikowski’s book represents a very different kind of scholarship—more theoretical, abstract—from the mainly experimental work found in our readings for this week. Nevertheless, in that book, the author sought to apply ideas from recent research in cognitive science to theorizing and analyzing music. I read the book as part of a reading group that I and my fellow music theory graduate students organized, so I had the opportunity to talk about the book with some of my peers. While I found some things that were stimulating, I have to admit that overall I and my fellow reading group members were unimpressed by the book. This I think was mostly because we didn’t feel that it had anything especially new to say: what we saw was some fairly “traditional” music analysis mixed with a few new terms borrowed from the cognitive sciences (“concept,” “category,” “conceptual blending”).

    I bring this up here because it raises an important issue that seems to crop up whenever I read or have a discussion about music cognition scholarship, having to do with the nature of the relationship between music theory and music cognition research. I often notice that the kinds of questions that cognitive scientists or psychologists ask about music are quite different from the ones that music theorists tend to ask about music. This, I think, is reflected in the difficulty I had with Zbikowski’s book, and, I would say also, with the research I read about this week. Put simply, I often find that the kinds of questions cognitive scientists and psychologists ask about music are not particularly interesting to me, a music theorist. This isn’t to say that I think this sort of work is unimportant and not worth pursuing, because it is the opposite of this that is true; rather, I think that there tends to be a disconnection between the goals of music cognition and psychology and those of music theory. Where, for example, Martens is interested in the question “at what level does a listener perceive the tactus of this piece of music,” I would be more interested in the question “at what level can a listener perceive the tactus of this piece of music.” That is, as a music theorist, I am less interested in defining perceptual ground basses and norms (thereby tacitly limiting what is possible) than in exploring and cultivating a variety of different ways of hearing musical phenomena (like the tactus of a piece of music). For example, in the version of the tactus-finding experiment that we took part in this week, I found myself wanting to try out different ways of hearing the pulses by focusing on different metrical levels or focusing on different parts in a polyrhythm.

    I also find that the results of some experiments in music cognition and psychology are not, from my perspective, especially exciting or illuminating. For example, the findings of Janata, Tomic, and Haberman, are, for me, too tentative or too much in accordance with what I tend to take for granted (e.g., that we are compelled to move when we listen to music and that we find doing so pleasurable). This, I would say, also accounts for my lukewarm reaction to reading Zbikowski: I was not convinced that the research he was reporting on could enhance music theory and analysis in interesting ways. I understand that not every experiment will lead to earth-shattering results, that there’s much more to it than that, and that it’s easy to be dismissive of an experiment that seems to have only told you what you already know when the real question is not that it is so but why it is so. What I feel it all comes down to is that much music cognition scholarship is really more about “cognition” than it is about “music,” meaning that it tells us more about how the brain works than about the music. As a music theorist, I consider my primary interest to be in the music, not in the brain, and I would expect that the primary interest of cognitive scientists and psychologists would naturally be in the brain, and not the music.

    This, I think, gets at the real issue: the differences between the aims and methods of scholars in the sciences versus those of scholars in the humanities. This was among the issues discussed in Huron’s article on “the new empiricism.” From my perspective, the field of music theory is aligned primarily with the humanities: the study of music is the study of an aspect of human culture, not the natural world. I would argue that music is not a natural phenomenon, but an artificial one (in the sense of being artifice), and therefore a human one: as I like to say, pieces of music do not grow on trees; they are made things. So, the people who make music their primary object of study—music theorists and musicologists, for example—fall mainly within the purview of the humanities. On the other hand, our psychological and physiological responses to music are, arguably, natural phenomena (though, I hasten to add, natural phenomena that are probably highly variable and subject to cultural conditioning), and so the people who make them their primary object of study—cognitive scientists and psychologists of music—fall mainly within the purview of the sciences. It should come as no surprise, then, that the goals of the two—music theorists and music psychologists—should be so different. One may lament that this is the case, and wish to see a greater integration of humanistic and scientific studies of music, but doing so may result in one aspect suppressing the other, and so I think it’s important that their independence be maintained.

    • I appreciate your commentary on the different aims of the sciences versus the humanities and how that shapes the questions and methods used to investigate them. There is one small point I would like to clarify though. You state: “Where, for example, Martens is interested in the question “at what level does a listener perceive the tactus of this piece of music,” I would be more interested in the question “at what level can a listener perceive the tactus of this piece of music.””

      The question you are more interested in is not at all outside the realm of psychology or the cognitive sciences. It would only require a different experimental design. You might use a focus group with who you would test possible hearings derived from analysis, you could have a study focused on a single style-specific expert musician, the task could involve free response and/or qualitative rather than quantitative or action-based responses. Empirical research is research that uses observation to derive knowledge. This is different from purely experimental research, which requires some form of manipulation by the experimenter.

    • Imri’s response to Simon


      I think you’ve articulated questions about the relationship between music theory and music cognition which I also find problematic. In particular, it seems to me that (and this is to be taken with more than a grain of salt given my knowledge of the field) at this point, most empirical experiments try to establish basic models with which one can “scientifically” prove what we already assume in practicing music theory/analysis. I guess this is unavoidable in a field that is relatively new. In that sense, I share your feeling that as a cultural construct, music is still better interpreted systematically with regards to its own history, forms and culture.
      On the other hand, there’s something incredibly exciting for me about the naïve/utopian notion of knowing “how things actually work”, or in other words, to be able to bridge certain leaps of faith which we tend to make in music theory, in the lack of sufficient knowledge of perception. Also, I’ve already probably mentioned all-too-often how problematic I find analyses which go against preference rules/perceptual categories and seem to stretch the degree of intellectual simulation involved in analysis unjustifiably; this is something that could really be criticized more productively using such knowledge.
      Finally, I find the possibility that composers will one day have sufficient knowledge of perception to write with utopian control over the effect of every note to be a nice ideal to strive for (even if there’s ultimately something totalitarian about it). Something of this discipline of psycho-acoustics is already present in Spectral music (especially in later Grisey), and I don’t see any reason why this Scientific/aesthetic junction should not stay productive for a while.

  6. Joseph Prestamo
    Current Trends in Music Theory
    Musical Cognition

    Where many music theorists might have decided to investigate the musical features that constitute the common notion of a “groove,” Janata, Tomic and Haberman set out to explore this phenomenon as a psychological construct. This was a fascinating idea to me, because my first inclination (one I would suspect would be shared with many other classically trained musicians) would’ve been to examine the rhythmic and harmonic features present in a sampling of different grooves to look for patterns and connections. But Janata and his colleagues are interested in the cognitive experience a listener has when engaging with a musical groove (specifically, what they refer to as sensorimotor coupling), which leads them in a very different direction.

    They begin by attempting to define a musical groove as a psychological construct, with several parameters in mind, such as: is the concept of a groove consistently represented in the listener’s mind? is it an attribute of music that can be perceived and consistently judged? is the listener appraising an attribute of the music or the subjective quality of his own sensorimotor interaction with the music? In order to do so, they employ a series of questions given to the participants of the experiments. I found the results of the initial surveys to be quite fascinating. Once again, I would have approached this subject from a music-theoretical perspective and my primary definition of a groove would have been dependent on a steady, underlying beat. However, the rhythmic component of a groove was in fourth place according to the survey results. The top three spots were given to completely experiential attributes: a) the groove depends on the extent to which the music makes you want to move, b) I like to listen to music that “grooves,” and c) the groove depends on the extent to which you feel you are a part of the music. I suppose this is an example of one way in which approaching the concept of a groove as a psychological construct enables Janata et al to consider aspects of the musical experience that most likely would not have made it into a music theory analysis. The idea that one’s enjoyment of a particular musical experience can be included as part of its very definition in this context is incredibly intriguing to me.

    Along similar lines, the article closes with a few rather provocative questions raised by implications of the experiment’s findings, regarding entrainment and social behavior. Janata and colleagues ask if it is possible that “high-groove music essentially serves as an invitation to join the group by virtue of inducing an urge to move along with the actions of the group?” Indeed, speaking of a groove as something that makes you feel like “you are a apart of the music,” or even the simple phrase “in the groove,” which was used throughout the article, seem to imply that the listener is becoming a part of something larger than himself. This makes intuitive sense; dancing is often a group activity, and a groove certainly does seem to carry a sort of unifying power. I am not sure how I would conduct a further study of these implications, but the topic does seem rich with possibilities.

    Martens set out to discover what strategies a listener might employ to choose a tactus for a piece of music. Specifically, he wanted to know whether this choice could be influenced by subdivision in the musical surface or simply by a tempo-based model. The methodology by which he investigates this seemed reasonable to me. At first, I wondered if measuring the participants’ taps would truly be a good indication of what they perceived to be the tactus. It seems possible that the participants could have simply been inclined to tap something other than what they believed to be the main pulse. For instance, if a particular subdivision catches the listeners attention and they tap along, does that really mean they believe it to be the tactus? However, after a closer read, I think the instructions to “tap at a steady and comfortable rate along with the musical excerpts” are a good prompt to let the music guide the listener’s intuitions toward selecting the tactus, rather than some other beat.

    Martens found that a group of listeners tended to tap along with a subdivided pulse (a group that he called the Deep Tappers), another group tended to follow the fastest consistent pulse (which he called Surface Tappers), and a third group used a combination of these two strategies (which he called the Variable Tappers). Martens takes the position that these strategies are fairly malleable, and I would agree that they are most likely shaped by our musical experiences, training and listening habits. When discussing how a composer might utilize these findings, Martens quotes David Huron, saying: “As we learn more about individual psychological differences, in the future it might be possible for composers to tailor works for particular subgroups or even individual listeners.” Predicting how an audience might react is indeed one of the most difficult puzzles with which a composer can grapple. Therefore, it stands to reason that information about how a group of people may interpret a musical work would be incredibly useful to a composer. However, because of the malleability of our listening experience, Martens seems to think that it would be difficult to actually accomplish this, and I would agree that it does not seem like a reliable prediction.

    A more interesting investigation, in my opinion, would be of the composer’s own musical intuition when selecting (or creating) a tactus. In my own music, I have noticed a tendency toward slow pacing or rate of change. This is not always a conscious choice, but usually an intuitive one that is manifest in slow tempo markings, long note values, or slow rates of harmonic change. Even when the tempo is fast and the note values are short, there often seems to be some type of a slow pacing still present. Perhaps, this is unrelated, but I also found myself agreeing with the “deeper” listeners choice of tactus in the musical examples discussed. Perhaps we cannot use these findings to always predict how a group of listeners will react (or even how we ourselves will always react), but learning more about my own motivations and musical intuitions has always proven to enrich my creative experience.

    • Hi Joey, thanks for your summary and commentary. I agree with you that it’s interesting to include psychological aspects of groove in the definition of “groove” and indeed it does seem to go against the grain of how we typically define things in music theory. However (and I think you agree) in this specific case, it seems rather reasonable. I believe the word “groove” itself arose among performers of popular styles like jazz, blues, and rock, and all performers are very concerned, consciously or subconsciously, with the idea of embodiment of the music and probably of the groove. Any definition of groove, to be satisfactory, ought to recognize the part that embodiment has to play in our conception of groove. Honestly, I’m not sure how strictly music theoretical the word “groove” truly is; I think it’s a word we use to conveniently describe this urge toward embodiment that surfaces in certain kinds of pieces more than others. What intrigued me more about that study was the idea of quantifying the groove in a “groove rating”! This seems very counterintuitive to me, again because it is not really about a quantifiable phenomenon, but a principle of embodiment, and it’s hard to say to what degree one embodies something.

      I think you’re proposing a very interesting question by wondering what the perceived tactus is for composers of their own pieces. One could ask several questions in this vein: how often does the composer’s tactus correspond with the beat unit that the composer has indicated in the time signature? Or, how often does the audience perceive a different tactus than the composer? What about the performers—how do their conceptions of tactus correspond to those of the composer and the audience members? I’m not sure what the broader applicability of such a study would be, but they are certainly interesting questions.

  7. Justin Reardon
    Current Trends in Music Theory
    Prof. Joseph Straus
    November 24, 2013

    In my response to the readings and experiments offered to us this week by Prof. Poudrier, I have chosen to focus on the Martens study, as well as on my own experience with the online experiment on spontaneous grouping administered through the MARL platform. Interestingly enough, the ideas discussed in the Martens article seemed to filter into my subsequent experience of the online experiments, as there is considerable overlap in the material studied in both cases.
    As described at the outset of the article, the primary goal of the Martens study is to attempt to model the listener’s choice of tactus, given a musical surface, via a tapping task. The conclusion, as stated in the beginning of the article and elaborated in the discussion at the end, is that the notion of tactus is ambiguous, in that there is no evidence that tapping behaviors in response to music demonstrate a global basis in tempo or the subdivision benefit. Instead, three types of listening strategies are identified by the study: the tapper that focuses on a subdivided pulse layer, the tapper that focuses on the fastest consistent pulse (FCP), and the tapper that combines a mixture of the above two approaches. In other words, if tactus is defined as the “main beat” chosen by the performer or listener (and not necessarily reflected in the choice of notation in the score), the same musical surface can elicit a number of responses in terms of perceived tactus. Ultimately, this seems to suggest a few basic ideas: that the “preferred tempo” range (as represented in the study by the peak region of a resonance curve) does not necessarily predict tactus choice, that the cumulative experience of the listener (i.e. musical training, listening habits, performance habits) has a significant role to play in tactus choice, and that the relative tempo of a given musical surface also has a considerable impact on tactus choice. In the body of the article, the author elaborates on the above suggestions.
    As the primary purpose of the study is to investigate the notion of tactus in real musical examples, the question posed by the study is whether or not the subdivision benefit is universally observed in listeners; given the subdivision benefit versus the tempo-based resonance model, the author wants to know which better accounts for the listener’s response. He therefore posits the hypothesis that, given music that features at least two consistent layers of pulse that are both above 33 bpm, the listener’s tactus choice will be a consistent pulse which can be divided by at least one other consistent pulse present in the excerpt (i.e. the listener’s tactus will not be the FCP, but rather a subdivision thereof); this is shown not to be true based on the study. In order to operationalize the initial question, the author gathers a group of stimuli and subjects, and builds an apparatus for the study that utilizes MAX/MSP and a plexiglass surface to record tapping responses; these responses are compared to a resonance curve, a set of hypothetical subjects, and ultimately are compared amongst themselves, which renders a distribution for listening strategy within the group of subjects. The primary measurable feature of the study is speed of consistent tapping, expressed in bpm.
    Though I am not well read in studies of music perception and cognition, the “operationalization” step of the study seemed generally sound and plausible to me, and offered results that were readily intelligible. Based on a hierarchical cluster analysis of the group of subjects, approximately half of the subjects are deemed “Surface” tappers (relying on the FCP predominantly in their response), whereas the other half of the subjects are deemed “Deep” tappers (relying on SP2), and “Variable” tappers (relying mostly on SP1). Furthermore, tempo seemed to have a significant effect on listening strategy: very slow excerpts encouraged more reliance on the FCP across the subjects, forcing the Deep tappers to behave more like Surface tappers (and even encouraging the Surface tappers to utilize an inferred pulse), and very fast excerpts encouraged heightened reliance on subdivision, forcing the Surface tappers to subdivide and the Deep tappers to shift to an even deeper level (here, the author notes that the behavior seems cyclic, in that the entire distribution of responses shifts down one level in terms of subdivision). In more basic terms, more extreme tempi seem to make listeners modify their natural tendencies, and tend to result in higher failure rates to properly synchronize with the musical surface (this is especially clear with the Surface tappers). Perhaps most interesting among the results are the suggestions that specific types of musical training influence listening strategy, and that it is the Variable tapper that is perhaps the most concerned with the specific rhythmic activity of the musical surface in choosing tactus (this type of tapper also taps more often with the notated tactus).
    In general, as would be the case with any study, the Martens study is limited by the various components of its methodology. First and foremost, it is limited by its choice of stimuli: the amount chosen (30), the duration of the excerpts (20 seconds), and the selection of passages with common meters and temporal regularity. The study could be expanded by incorporating more excerpts, and of longer durations, and is unable to discuss passages that feature significant tempo changes, changes in meter, irregular meter, or metric modulation (all of these metric issues could warrant studies of their own I suppose). It is also limited by the number of subjects chosen—a larger group could be studied in the future. The apparatus used only accounts for a tapping response (the movement of one body part); other forms of movement in response to music could be studied in the future (multiple body parts). Perhaps most importantly, given the diversity of musical excerpts chosen (here, I am speculating, since I am not privy to the exact excerpts), the study is not able to isolate the particular features of the musical surface, beyond tempo, that might contribute to listening strategy. Being a composition student myself, I certainly find the study to be relevant to my own work; the findings of the study would encourage me to treat issues of the particulars of the musical surface as well as the underlying structure (particularly where subdivision and grouping is concerned) with equal attention, as there will always be several types of listener in any given audience.
    The Martens study connects most directly to the “Spontaneous Grouping” experiment that is featured on the MARL platform. While completing this experiment, I found myself thinking that I would definitely be considered a Deep listener. Given slow tempi, and in particular the alternation of tones separated by a semitone, I selected smaller groups (2 or 3), but given faster tempi and neutrality of pitch, I consistently found myself projecting larger groups onto the beats (4, 8, 12), even though I was conscious of the fact that they were in essence the same two- or three-group patterns sped up. Furthermore, I had a preference for multiples of two rather than three, suggesting that I had a desire to project symmetrical groupings (I seemed to impose 4 and 8 on simple two-beat groupings, and only selected 6 or 12 to project compound meter if a fast three-beat grouping was present, with a preference for 12). I am a composer and pianist, and I have also conducted (in addition to the fact that I recently began to take dance), so I suppose that my status as a Deep tapper would not come as a surprise to Martens, given his commentary on musical training and tactus choice. I will be very curious to read about the observations gathered through this experiment, as well as the results of the other two experiments.

    • I appreciate your highlighting of Martens’s finding of an interaction between tempo and listener’s preferred, in particular how extreme tempi seem to push listeners to change their tapping behavior. You also make several interesting suggestions for further study of this phenomenon.

      One note: It would be entirely possible to do post hoc exploratory analysis of the musical features that seem to correlate with various participants’ responses. Psychologists don’t tend to do this for several reasons, one being that no psychology journal would publish the results of these analyses, and another because most of them don’t have the training required to do so. But I am confident that there might be researchers out there that would be very interested in collaborating with a music theorist who would engage in this kind of analysis, if only because it would be likely to produce observations that can be passed through the “question-theory-conjecture-hypothesis” grinder.

      In my Poudrier & Repp (2013) study, I gathered enough courage to include one small attempt at post hoc music analysis and one of the reviewers (M. R. Jones, a giant in experimental psychology) expressed her appreciation of this speculative exercise exactly in this way, i.e., as interesting information that could be used for further experimental research.

  8. While reading Martens’ article, I began questioning the validity of the connection between what one listener hears/feels as the tactus and what she ‘taps’ on the 8” by 11” surface. To me it seemed that a few things were being taken for granted. From reading the article on the concept of groove (Janata, Tomic, and Haberman) one easily understands that groove as a feeling manifests itself through bodily movements. I understood in that article that there is something especially bodily holistic about groove (compared to tactus for example.) Hence groove pushes us to dance for example – which seems as a most natural observation. On the other hand, we cannot find the same logical/empirical strength in the association between tactus and tapping. Reading Martens’ article made me admire the highly quantifiable nature of tapping, that fascinating way of measuring perception. Perception being a very debated and complex matter, giving it a number allows both a statistical analysis and an ‘undeniable’ definition. But I was also wondering how tapping truly represents our internal perception of the tactus. Unlike the feeling of groove (where subjects seemed to naturally be tempted to move or dance), the concept of tactus is in my opinion a more mysterious one. It involves both grouping and prioritizing. It also seems connected to the innermost effects music has on our psyche. For some reason, when I think of Leo Tolstoi’s ideas on music (from The Kreutzer Sonata for example) and its powerful effects on the listener’s emotional state, the idea of tactus seems more appropriate than the feeling of groove. It’s as if a felt tactus was an ‘internalized’ groove – and vice-versa. Also, perhaps precisely because it is usually un-manifested in physical ways, our perception of tactus may become more complex and become much harder to understand and quantify. Unlike the feeling of groove, the perceived tactus does not necessarily push us to move. In other words, the feeling of tactus does not push us to tap our fingers. Perhaps it would have been of great use to make a ‘Tactus attribute ratings’ survey – much like the one made with the idea of groove on page 57 of Janata/Tomic/Haberman article – so that we could have a slight idea on how most people perceive tactus.
    The human body is infinitely complex, but what is even more intricate is the relationship between mind and body. For example, if Martens’ subjects were asked to tap their feet instead of their fingers, would the results be different? I would believe so. What if the subjects were asked to ‘bounce’ their heads to the music? The results would be even more different (‘Surface’ tappers would probably be a most tiny group!) Also, one has to take into account the ‘mirror’ motion – not what the mind tells the body to do (in Martens’ case, to tap) but rather how the body influences our perception. We can witness that, for example, with breathing. By controlling our breath (its speed and width) we can strongly influence our emotional state, hence our thoughts and perceptive qualities. In Martens’ experiment, tapping may have had a stronger influence on perception than we would at first imagine. For example, the muscles used in the tapping motion may have certain characteristics that make faster motions easier. This especially seems to be the case with the index and third finger (tapping faster with the thumb for example is harder.) Hence, I would have liked to see the ‘finger’ variable taken into account. Also, the angle of the hand to the forehand is of crucial value, since fingers move more easily under a certain optimal angle.
    Once we start working with the human body, it is my belief that we simply cannot isolate certain parts of our body and ignore other central parts and organs. For example, in making the distinction between the ‘Surface,’ ‘Variable,’ and ‘Deep’ listeners, various crucial variables were not taken into account: Their heart beat at the moment of the experiment, their breathing pace, their blood sugar, and the volume of their lungs (which influences the breathing pace) among many, many other possible variables which can be inspired from any serious athlete’s body check. These variables can be surprisingly changeable within a single person, depending on its physical condition (nutrition, sleep, emotional state, etc.)
    Concerning the Janata/Tomic/Haberman article on the feeling of groove, I found that the authors’ initial focus on the actual word groove was helpful, although rich with contradictory material. By starting the study with a single word (concept), the authors allowed the subjects to experience a multitude of feelings – precisely everybody has a relatively different concept of what the word groove means. What I would propose to do, as a possible complement to this article, is to flip the multiplicity around: In other words, instead of having one word and a multitude of feelings and personal meanings, we could have a multitude of words and a single feeling. For example, the authors could have defined a certain physiological response to music (which the authors would call groove, and which would be measured with precise instruments and using various biological variables) and analyze to what degree do subjects experience a likely reaction to music. They would of course all give different names to that particular feeling (according to their social and cultural backgrounds) but the linguistic aspect of the study would only appear at the very end. Vocabulary being a very complex matter, by refusing to start our study with a focus on the meanings of the word groove, we eliminate the complexity that lies behind this linguistic and cultural phenomenon. The definition of the word groove would have appeared at the very end of the study – as a natural consequence of a series of physiological observations.
    Obviously, as soon as we start working with the human body and mind, the amount of variables that have to be taken into account is exponentially multiplied. Nevertheless, this approach is essential for some sort of perceptual objectivity to appear. Nowadays, it is the A.I. (artificial intelligence) computer programmer’s dream to recreate a human machine. And although I believe we will never create a human being electronically, its five-sensed perceptions can be measured, hence reproduced. This means that no matter how complex the perceptual variables are, we may eventually be able to create a research model that closely dissects the way both our mind and body respond to music.

    • I appreciate your emphasis on the role of embodiment in our experience of musical rhythm. Based on reported experience from people in a wide range of musical cultures and historical periods, there indeed seems to be a strong connection between movement and musical time perception. And in fact, more recent experimental research has been able to identify at least one important contributor to this phenomenon, i.e., the vestibular system. The most popular theoretical model of beat perception right now, i.e., the Dynamic Attentional Model of M. R. Jones, is also supported by a neural oscillator model that depends on the entrainment of internal neural processes to external auditory signals, another form of embodiment (which has been observed in babies that are only a couple of days old).

      I can see how finger tapping might seem to be less convincing evidence of an urge to move to music than dancing, but it has proven to be an extremely reliable manifestation of induced pulse. The up-down movement seems to be widely spontaneous (e.g., think of how often you’ve witness someone bobbing their head or their feet to the beat), and tapping a finger requires minimal effort. It is true that if the spontaneous sensorimotor synchronization is experienced or expressed through one moving their head, to be required to express it with one’s finger instead will require a physical mapping or transference of sorts, but it has not seem to have a big effect on findings so far. Nonetheless, such findings will be only made stronger by using a slightly different operationalization involving different modes of response.

    • Dear Nikita,
      Thank you for your response to the readings that we did for this week on perception and cognition. I found these studies to be very engaging, and enjoyed participating in the online experiments as well.
      I understand your hesitance to accept the tapping response, produced by one finger on a surface, as the primary measurable phenomenon of the Martens study; here too, I immediately thought of the Janata study and the fact that the research time involved in that study attempted to measure a broader set of bodily movements in response to musical stimuli. Of course, the tapping measure is much more economical, but perhaps the Martens study overlooks a broader sense of embodiment with regards to tactus choice; on the other hand, from an intuitive point of view, tapping seems to me to be a fairly reliable indicator of the internalized sense of pulse. As you suggest, I think it would be interest to investigate whether or not different body parts “count” in different ways; perhaps the lower extremities would tend towards deep counting (as well as the head), and the upper extremities and fingers would tend towards surface counting. I think it could also be interesting to monitor automatic nervous system responses while a subject is asked to find and tap a pulse; there are certainly studies out there that use such measures as their primary mode of quantification.
      With regards to the Janata study, I found the exploration of the term “groove” to be a helpful beginning, and a way for the authors not to make assumptions about what the concept of groove entails; their goal is to connect bodily movement with perceived “groove level” in musical excerpts, and this couldn’t really be done without a preliminary exploration of that which makes a musical passage “groovy.”
      Finally, a quick question: you are a violinist (player of an instrument with a relatively high range, which is often employed for melodic purposes)—did you find yourself to be a “Surafce” or “Deep” tapper in completing the Spontaneous Grouping experiment? Or something in between?

  9. Javor Bracic
    Current Trends in Music Theory
    Prof. Joseph Straus

    I found this week’s reading highly interesting, as I am deeply fascinated by concepts of music perception and cognition. In my response, I will focus on the study of musical groove conducted by Janata et al, and briefly share some thoughts on the online experiments I took.
    The goal of Janata’s study was to introduce, observe and quantify the phenomenon of “groove” that some music reportedly posesses. The interest in the subject, primarily from a psychological standpoint, was not to construct the concept music-theoretically, but to analyze its reported perception in a group of young adults.
    The participants were asked to define the idea of groove and rate musical examples on a scale of “grooviness”. In a second study, other individuals were asked to tap along musical examples in various ways and report the levels of enjoyment and difficulty at each groove category.
    According to expectation, groove was found to be bound to the idea of bodily movement, and of a person being compelled to move by the music. The most groovy music was found to be music with a faster tempo, and of a specific genre, namely soul/R&B. The authors explain this result with the presence of a strong sense of the beat and accentuated, danceable rhythms in soul/R&B music. After tapping to the music, the subjects of the second study reported greater ease with groovy music, as expected. I was left wondering, though, what was it exactly in the R&B rhythms that made this music more groovy, to which I will return later.
    I found especially thought-provoking the relation of groove to Csikszentmihalyi’s concept of flow. The state of flow, according to Janata’s interpretation of Csikszentmihaly, is achieved when a difficult task is performed well. Janata posits that groove and flow, even if related, differ slightly, and he attributes this to the dip in perceived difficulty for tapping to groovy music. If groove and flow are intuitively related, why does very groovy music seem easier to tap to than medium-groovy music? Janata’s conclusion is that they are not absolutely correlated.
    This is where I disagree with the authors. Can we really trust the self-observed measure of difficulty? I strongly believe flow and groove are two faces of the same coin. Therefore I propose that the reported measure of difficulty is not objective enough and I would be so bold as to attempt to back my assertion with elements of music theory.
    Firstly, as one of the conditions for achieving flow, Csikszentmihaly describes the balance between the challenges of an activity and the skills of the person engaging in it. “A piece of music that is too simple relative to one’s listening skills will be boring, while music that is too complex will be frustrating. Enjoyment appears at the boundary between boredom and anxiety, when the challenges are just balanced with the person’s capacity to act.” (Csikszentmihaly, 1991) Therefore flow is not only about having a difficult task at hand – it is about matching the difficulty to the curent skills.
    Secondly, in trying to figure out what was it about the music itself that made it groovy or not, I discovered a possible trend: Non-groovy music fell into two categories: rhythmically straight-forward or rhythmically complex, whereas groovy music hit the sweet spot in between – it contained a predictable, simple and reliable pattern, as well as an off-beat, syncopated element that was more challenging to follow. Other elements seemed to play a role too, such as rubato – which was unfavorable, as it made the pattern less predictable, therefore more challenging to follow.
    Let us consider some examples. Superstition, which was rated highest on the groove scale, contained a steady regular beat, and a relatively irregular, syncopated accompaniment and melody. So did Flashlight, Dip It Low and Yeah! This is especially clear in Yeah! where an extremely simple regular beat is juxtaposed to a two-note melodic phrase that changes from being beginning-accented to being end-accented – I believe we feel tension between the melody and the rhythm because of this constant change, and that tension is what sets us into motion, and enables us to groove. Goodies had a similar but inverted story – the melody was perfectly regular and predictable, but the bass revolved around an irregular 3+2+3 beat, so that even the regular melody was experienced as bursting with tension in contrast to the bass – which was again acting as momentum for movement.
    On the other hand, songs that were too straight-forward, like Down by the Sally Gardens, probably did not challenge enough – tapping was indeed simple – but there was nothing in the music that compelled us to tap, therefore it took extra effort to continue tapping with each new tap. Same goes for Flandyke Shore, one of the least groovy songs in the list, where the presence of rubato and ritardando further slows down the beat and adds considerable effort to tapping. This is certainly nothing to take away from the aesthetic value of the music, on the contrary! Next, I wondered what made TFS by Herbie Hancock the least groovy of all the groovy soul/R&B category. It exhibited a steady beat, and some irregular patterns too, but as a whole it was too unstable and unpredictable. The Girl from Ipanema was a very interesting example indeed – I needed to actually write down the rhythm as in a dictation in order to even wrap my head around it! Here both the bass and the melody play irregular rhythms that do not seem to match each other at all, which creates a frustratingly difficult tapping task – and by the way, adds a wonderful elusive ambiguity to the music.
    Lastly, I would freely ask if there is a possibility that the perceived difficulty reported by the subjects was blurred by the enjoyment experienced due to the flow resulting in the balance of challenges and skills. According to Csikszentmihaly, flow (also called the optimal experience) is one of the greatest sources of happiness and fulfillment in human life. Therefore it is easy for me to imagine that the experience of flow arising from tapping to music that had the exact right amount of challenging elements was the source of greatest enjoyment – and that it felt easier than tapping to a slow and straight-forward meter.
    As a performer myself, I can testify that some difficult polyrhythmic passages (i.e. one hand playing 3, the other 5 beats per measure) can only be played correctly if a certain level of entrainment and “autopilot” has been achieved – and then it suddenly feels easy!
    All of this is not an attempt to overthrow the conclusions of the study – rather I’d like to see future studies that would prove or disprove my hypothesis. I could well imagine identifying musical examples from the classical period which would exhibit similar rhythmic complexity patterns and conducting a study that would strive to show a correlation between these patterns and the experience of excitement and enjoyment – flow in music.

    I thoroughly enjoyed taking part in all of the online experiments. At the beginning of the “Perceived Emotions” experiment I was tired, stressed and anxious due to the amount of work in front of me on a Sunday afternoon. At the end, I felt somehow content and somewhat more relaxed, even if a little bored (listening to 40 repeating examples in one sitting was a little tough, but it was great fuel for my procrastination!) By the end of the experiment, I felt like my judgement started to wane though, and I believe I even made some mistakes on the familiarity measure for the same pieces. (which might be taken as proof that a piece in a slower or faster tempo may seem a little less familiar than when played in the original…?)
    In the “Spontaneous Grouping” experiment I basically only heard 1, 2 and 3 groups. I did sometimes group them in my mind into 4,6,8 and even 12 and 16 groups at times, but these sensations were too fleeting and flexible within one example in order to be able to put my finger on them. And so I gained a valuable insight: I believe it was my musical training that impeded me from making instinctive judgments. I listened carefully for any sign that the groups were different, and when I found none, I made a conscious decision to select the most censervative and correct choice. It felt like choosing a larger tactus was a matter of choice for me, just like in an optical illusion – you can with some effort switch from seeing an image as a vase or as two faces, and you know both are there, even if you can see only one at a time. Therefore I would conduct this experiment primarily with non-musicians who might be more prone to going with their gut feeling.

    • Tom Johnson
      Response to Javor

      Dear Javor,
      Thanks for your summary of and reaction to the Janata et al. study. As a study by psychologists of something music theoretical, rather than the music theoretical approach of Martens to something more psychological, the Janata experiment seemed a bit contrived to me in its problematization of the word “groove.” They describe groove as “the pleasing state in which the creation of music becomes seemingly effortless” and in which participants generally exhibit some sort of sensorimotor reaction. But, when they choose undergraduate subjects in ~2010 for their experiment, Janata et al. bias their results to a meaning of groove already firmly established. Really, to understand what “groove” means to a ~20 year old college student today, we need look no farther than the common topicological associations of the word groove that a college student would have. It is a relatively antiquated word at this point in popular culture and in college-aged lexicon, as is easily seen by a quick Google image search of the word, where 1960s hippie and 1970s funk depictions abound. So, is it really surprising at all that funky soul and R&B songs from the 1970s scored highest on the groove scale since they are essentially the reference point of grooviness for the subjects at hand? I know this doesn’t engage with all of the details of the study, but I couldn’t really get past this sort of self-evident biasing and seemingly circular logic.

      Sorry to go off on a tangent like this, Javor, instead of responding directly to your shrewd assessment of the relationship between groove and flow. I agree with you, especially in regard to the issue of difficulty in flow. Personally, I find the asymmetrical 5/4 groove of “Take Five” to be especially groovy, but it is perhaps only because I have played it and am so familiar that the relative difficulty is so low to me.

      It is interesting to me that your self-conscious choices resulted in groupings of relatively small groups in the “spontaneous grouping” experiment. Personally, I found myself trying to see if I could feasibly make larger groups, or if indeed groups of 5 or 9 made any sense (which they didn’t). I agree that, as musicians currently reading music cognition studies in a music theory class, we were likely metatheorizing a bit too much as we took these little tests. It would be interesting to see how our class’s results would relate to a less biased group’s!

    • You ask: “Can we really trust the self-observed measure of difficulty?”

      Self-reported measures indeed require to be used with caution. Psychologists are especially sensitive to this issue. This is why such measures are usually accompanied by other measures that can be used to confirm or problematize the findings. In the study by Janata and colleagues, the familiarity measure is secondary to other measures. A follow-up experiment could include a more controlled manipulation of familiarity, for example, by hacing participants listen to previously unfamiliar music for a given amount of time and then have them rate their “grooviness.” There are multiple ways of doing this. The bottom-line is that studies that involve some form of training are very costly and even then, you can’t trust all participants to actually do what they are supposed to do!

    • It would indeed be interesting to investigate what is the relationship between “groove” and “flow.” One would need to begin this investigation by formalizing the concept of “flow” in some way, and it would be risky to base such definition on a single author. This might be avoided by beginning with some form of exploratory study using surveys or interviews of musicians.

    • Dear Javor,

      Thank you for your summary and critique of the Janata article. I did not read it as closely as the Martens, so it is nice to have it laid out so clearly. In particular, I really appreciated your encapsulation of the concept of flow. I have been reflecting on how that concept applies to pretty much all facets of living, but in particular how specific sensation of “flowing” with music when playing music is something uniquely exhilarating—that feeling you get when your fingers are on auto-pilot and you are free to make real expressive choices.

      I share your concerns about the risk of bias in asking a bunch of undergrads to rate songs according to how groovy they are. I think Tom is right to point out that in the year 2013, the term “groovy” is something of an anachronism, and that for today’s youth it primarily calls to mind images of Austin Powers, the 70s, and the musical styles of that era. I am not overly familiar with R&B, funk, or many of the other songs that were rated, but I suspect that in addition to rhythm, what listeners are responding to when asked to rank the “groove” has a lot to do with how the rhythm is bound up with the timbre, in particular, the punchiness of the electric bass. Other musical parameters also probably play a factor, such as harmonic vocabulary (the use V9 chords, for example).

      I also had some frustration with the online rhythmic grouping test for partly the same reasons you describe. I think I misunderstood the instructions for the first part of the test (I hate mornings), because I was just counting things as being in two or in three, when really I could have been easily counting in larger groups. At some point, I got caught up with a string of sixteenths that I clearly heard in 4/4 and remembered that you were supposed to count that as a higher group, so I wasn’t sure how to proceed, since the counter only went up to 12. I wasn’t sure what to do with the test after that. I guess I put 4 for that one? When the examples started, it was always easy to group the notes as subdivisions of 2 or 3 (because of accent, pitch, etc.). After that, one could project almost any higher grouping. For me, this mostly ended up being eighths, triplet eighths, or sixteenths in 4/4. Apparently that’s my first-level default when given a string of repeating rhythms. It’s hip to be square, right?


Comments are closed.