Attention, Entrainment, & Sensorimotor Synchronization

Mari Ries Jones’s “Dynamic Attentional Theory” (DAT) has become the predominant theoretical framework used in the investigation of beat and meter perception in psychological research (for a description of the research conducted at her Ohio State University lab, see; the website includes a few PowerPoint presentations by Jones). While this theory was first presented in her article “Time, Our Lost Dimension: Toward a New Theory of Perception, Attention, and Memory” (Jones, 1976), it was the marriage of DAT with Ed Large’s oscillator model (Large & Jones, 1999) that “sealed the deal” for metric entrainment. Up to that time, most studies took a “goodness-of-fit” approach, often relying on some form of internal clock model (e.g., Povel & Essens, 1985; Parncutt’s 1994 “perceptual model of pulse salience”).

One of the most compelling aspect of the metric entrainment model of beat perception is that it is a powerful explanatory model that not only makes sense intuitively (“moving along with the music” is one of the most basic musical behavior, and entrainment is an observable phenomenon in the natural environment), but is also compatible with sophisticated theories about the communication of musical meaning (e.g., Meyer’s and Narmour’s expectation-based theories). Thus, it is not surprising that it has been embraced by psychologically-minded music theories (as represented by London’s 2004 “Hearing in Time”). The concept of entrainment has also caught the attention of several ethnomusicologists whose research had been focused on performance and embodiment (e.g., Clayton’s study of Indian music).

Review Large and Jones’s (1999) theoretical exposition and the experimental study that follows. What aspects of this work do you find most/least compelling? How are the findings “consistent with” the hypotheses derived from this theory? What are some of the limitations of the experimental design? Then, skim through Clayton and colleagues (2005). How is the concept of entrainment defined here, and how does it conform/contrast with Large and Jones’s definition? What are some of applications of the concept to music research as envisioned by Clayton and colleagues? Thinking back to Large and Jones’s experimental design, can you imagine an alternative (or complementary) experimental design that might address some of the limitations you (or the researchers) identified?

Write a preliminary response by Tuesday, September 24, 11:59 PM; the response should focus on at least one of these questions, and connect the two articles.


14 thoughts on “Attention, Entrainment, & Sensorimotor Synchronization

  1. I was struck by the seemingly counter-intuitive findings from experiment 1 discussed in the preamble to experiment 2: namely, that performance improved as a to-be-detected time interval departed from an expected point of time (large time changes). This seems to run contrary to the idea that performance should be best at moments of maximum attentional energy (i.e., moments that align with our expectations).

    The difference, as I understand it, is as follows: large deviations are more likely to cause a shift in our perception of the time structure, whereas deviations that occur at maximum points of energy attention are absorbed into the existing time structure. And so the experimental results seem to describe two different phenomena. But what is the relationship between the two? What does the spectrum of deviation look like in terms of our perception? In other words, when is a violation no longer normalizable within the existing time structure? And also, is there a buffer zone in which we don’t readily perceive variation at all, because out attentional energy is low?

    Another small note on performance: it would be interesting to study how performers place temporal variations in their performances. Do they intuitively situate expressive “violations” at moments that align with high points of attentional energy?

    • As far as your last question, I would point to studies by Clarke and Repp. The short answer is that performers show systematic as well as random patterns of variation. Within the systematic patterns, some are shared by most, and these pertain to musical structure: performers make events at the end of groups longer and those at the beginning shorter. Other systematic patterns are more individual or shared by a smaller number of performers. Repp’s series of studies of Chopin’s Op. 10, no. 3 provides a detailed analysis of a large number of performances and identifies several timing (as well as intensity) “profiles.”

  2. Perhaps what I find most compelling about Jones and Large (1999) is the big question, “How do people attend to events that change over time?” and the article’s attempt to find a rather uncompromising solution.
    By positing a model of oscillator(s), Jones and Large construct a theory of dynamic attention that creates expectancies for future events and uses these events to drive (and adapt) the oscillators. Jones has effectively created a hermeneutic model, in the term’s original sense (Schleiermacher’s); the model “shuttles” between events and oscillator(s), using events to develop the oscillator over time, such that the oscillator can more accurately predict the events in time. I particularly like the idea that entrainment relies on oscillators that train both to events and one another in deference to the different events. The introduction of coupling (the synchrony attractor) was an intuitive addition that greatly increases the model’s accuracy (ditto the phase coupling adjustments)—although I do not understand how coupling strength [the scalar between 0 and 2] is derived. (Is it just the value which yields the highest phase coordination?) Similarly, I found the reconstruction of a theory of expectation as a theory of attentional energy deeply intriguing. I think it only natural that predictive power changes over time (due to differing musical structure), and I appreciate the authors’ attempt to model that variation.

    Given my positive outlook on the efforts of Jones and Large, I might ask why Clayton et al did not import any of the sophisticated models of entrainment into their case studies. In an article that seems to advocate for the advantages of an explicitly oscillatory model, any use of such a model was noticeably absent from the three cases studies (intended to promote modeling entrainment—presumably through oscillation). In its place was a demonstration of the utility of some sort of sequential timing model. I believe that the variation in the lengths of the “long” beats of the diatonic timeline could be accurately modeled by Jones and Large’s two oscillator model. While I can’t really fault the analysis in the case studies, I do feel that Clayton et al missed an enormous opportunity.

    *One note. I was taken aback by the lack of discussion when Clayton “detrended” her data. While it is certainly necessary to adjust for variation in tempo when doing analyses such as these, I would not be so quick to adopt this procedure. In this instance, Clayton has essentially taken a data set, stuck a best fit regression to it, found that her data fit it better than the simple average (as will [almost] always be the case),and then discarded her un-manipulated data. This is not something to be done lightly. – Maybe this move is fully normal, and I’m just not accustomed to this literature.

  3. In reading Large and Jones, I found that a lot of the theoretical exposition landed somewhat over my head, and though I think I have a handle on the main ideas of the paper, I’m sure that I’ve missed a lot of the finer points. That being said, there were a few moments that caught my interest. The fact that the participants in experiment 1 were better at detecting late changes than early changes was notable. These results are not predicted by the model, which was pretty accurate about predicting everything else, and so I wonder where this effect arises from. Why should we notice more easily an absence of an expected note rather than an unexpected presence of one?

    On another note, as I was reading this paper, I couldn’t help but think of our discussion last week (an inclination then reinforced by the Clayton reading), about the difference between preference for isochronous and nonisochronous rhythm. This model seems, to me at least, to be dependent, or at least based, on isochronous preference, and so I wonder if there would need to be changes made to adapt it to a nonisochronous preference, and if so, what those changes mean for a difference in the greater concept of entrainment in these different cultures.

    I will also say that I would be interested in studies exploring the aspect of entrainment that is the ability to “tune in” to changes and adapt, rather than just detect these changes. The model accounts for this, but from what I understood there was no behavioral evidence for it sought out by these experiments.

    In reading the Clayton, my personal interest was piqued by the discussion of music therapy. This was actually something that came to mind as I was reading Large and Jones, as the subject of entrainment led my mind to the repetitive behaviors of autism and how the concept might be applied to an exploration of these behaviors. It was fascinating to see that there was some research into this relationship.

    • For the purpose of our class, you don’t need to understand all the finer details. You make some excellent points, and I particularly appreciate the fact that your relate some of these to the materials we covered last week.

      The contrast in performance between “early” or “late” events, i.e., the greater salience of the “absence of an expected” tone vs. the “presence of an unexpected one,” might also be related to the greater attention at expected beat locations.

  4. As Dr. Poudrier wrote, the basic concepts animating Jones and Large’s Dynamic Attentional Theory [DAT] meet little resistance from one’s intuition, e.g., that ‘temporal acuity’ should be much improved in tightly regular event sequences. Clayton et al. also suggest some avenues for research that seem natural by virtue of their apparent honesty to the biological grounded-ness of making music. But despite the obvious appeal of the attentional theory, I am left with some big questions and/or misunderstandings.

    Clayton et al.’s primary project is to expand the range of applicability of ‘entrainment’ as a useful concept in research, particularly in ethnographic work. Yet by widening this range, it appears they must also broaden the definition: “Entrainment describes a process whereby two rhythmic processes interact with each other in such a way that they adjust towards and eventually ‘lock in’ to a common phase and/or periodicity.” Clayton et al.’s definition is not far from the meaning forwarded by Large and Jones, but neither is it exactly the same. Nor does whatever distance there may be necessarily misrepresent the latter’s work. Recall, Large and Jones were perfectly candid of their embrace for general applicability outside of an obvious cognition of music context, even suggesting that their theory could feed back into visual-minded research, which constituted one of their own theory-building sources. Clayton et al.’s usage is significantly different in one way: their appropriation and expansion of ‘entrainment’ seems to be motivated enthusiasm for a novel worldview, another research lens and a new angle. That is, ‘entrainment’ broadens from a specific model or theory into a philosophy of sorts.

    But then, I could be wrong about this comparison. I admit to having had some trouble following the meat of the Jones and Large exposition. Afterwards, I still wonder about the specifics of the model’s interaction with rhythmically regular music. Does it really entrain a meter? (This is more clearly suggested in the affirmative in Clayton et al.) I get the sense that the answer is yes, given ‘multiple oscillations’. Theoretically, if multiple oscillations are attuned to multiple pulses, then presumably the model (Clayton et al. calls it: the ‘entrainment model’, if they even intend the equivalence) models ‘meter perception’. But my intuition still has trouble with the idea that one ‘entrains’ meter as it also does with the implication (perhaps wrongly projected by me) that it is more or less an autonomous process, which only require some energy in attention. It seems much more plausible to the same intuition that one ‘entrains’ a pulse, and that only after mental ‘imaging’, representation, and/or conscious reflection does one decide and conceive of meter, the system of pulses and their relationships. My perhaps poor understanding of the DAT in terms of the Large and Jones articles becomes exacerbated with Clayton et al. In the language of the latter, two bodies can ‘entrain’ with one another, e.g., ‘interpersonal synchrony’. But is that strictly possible according to the model? As I understand it, a host entrains with a perception, it’s processing of an external stimuli, which necessarily means that the host does not ‘entrain’ with another directly, but through a processed projection of the other. So I don’t quiet understand where Clayton et al. has gone other than to say what I already have: that they use the term much more broadly, if not metaphorically.

    -S P G

  5. I found the Large et al (2009) paper’s introduction to be a bit confusing with the interactions of the experiments. It seemed to me after reading the introduction that temporal relationships exist in the real world in daily interactions, but that when this “temporal structure” changes (as it often does) observers still tend to see these changes as stable, and want to make them become this way. However, it wasn’t until I skimmed the Clayton article and then went back that it all made sense: the “oscillations” grow to match each other rather than remaining distinct, as shown in biological life patterns as well as sociological patterns.

    I found the Clayton paper much easier to understand, especially regarding the application to study outside the musical realm. As Greta mentioned, it talks about the social nature of entrainment, and how the ability to “entrain” oneself with other people makes us able to socially connect with one another, which is especially important when considering diseases like Autism Spectrum Disorder, or even PTSD, where communication is a challenge. (The other application that I thought was interesting, though not particularly musical, was one related to social psychology. Perhaps we use the phrase “history repeats itself” because we are trying to make historical events fit into the molds of past events.)

    Once I had a better grasp on this, I went back to the Large paper. There, I stumbled again, and had some questions. First, in Experiment 1, this experiment seemed to show that this hypothesis may be proven because of the ability for even non-musicians (psychology students without a great deal of musical background) were able to better make the isochronous connections rather than the non-isochronous patterns. However, the graphs seem to show that both groups performed equally well on both sets of both tasks. How are those graphs meant to be interpreted? Clearly, I just am not understanding this.

    The one non-musical similarity that I thought was interesting between the two was in regards to a potentially visual attention perspective. As the Large paper mentions, we don’t know a great deal about visual perception/attention because the biology of physically seeing, plus the brain’s incredible task of weeding out unwanted stimuli at any given time are too intertwined to be separated. However, the Large paper seems to say that this, too, could be a form of entrainment in further study.

    I hope to figure out in a little more depth what exactly was going on in the Large study; I highly doubt that I have an accurate representation of what exactly they were trying to show their participants.

  6. Large and Jones’s model is elegant in its parsimony for explaining multiple facets of temporal perception including, but not limited to, musical, speech, and even biological rhythms. The standpoint of attention allows for a more biological impetus behind musical temporal faculties, thus removing specialization for the latter. This allows for a much more flexible model that can not only integrates non-musical faculties, but allows explication and integration for wide variety of temporal faculties. The latter is seen in continued research including metrical perception, musical and motor performance, as well as suggestions for polymetric and polyrhythmic perception (cf. work by Jones, Large, Pfordresher, McAuley, et al. and even myself!).

    The model is simple in that it involves three main state variables: phase, period, and focus. These each correspond with temporal faculties, which are clearly stated in the model but could be extended further. Thus, “phase is coordinated with the external rhythm, period approximates a pattern’s mean IOI, and focus reflects synchronization quality” (p. 132). This systematization allows for simple variable isolation for experimentation (as can be seen in the following experimentation), but also explains integration of the separate parameters.

  7. The Large and Jones article outlines a compelling and practical question. Why do we develop expectation and percept of rhythm even in the absence of a fixed temporal structure? The study in figure 1 reinforces the data behind this assertion, showing that in performance practice, the temporal structure is in fact varied from the strict rhythmic pattern outlined in the music when performed by a computer. Perhaps Large and Jones said it best: “An aspect of this puzzle then, is that people seem capable of comprehending a stable underlying temporal structure while at the same time responding meaningfully and adaptively to changing aspects of the time pattern.”

    In Clayton 2005, the concept of entrainment is much more clearly defined and more practical examples are given. De Mairan demonstrated in 1729 the propensity of plants to “photic entrainment”. The ubiquity of entrainment, from humans to single cell organisms is made apparent. This makes a much more practical argument in outlining and explaining the human capacity of rhythmic cognition. A plethora of biological processes serve to reinforce this assertion that we are, in the words of Pittendrigh, “a loosely coupled population of oscillators”. The concept of self entrainment is made evident in my own personal experience. When I run many miles, my heart begins to pump more quickly in order to move much needed oxygen to the muscles in my body. I begin to sense my heart rate without touching a pulse point. Often times, my breathing becomes synchronized to my pulse rate.

    Clayton is much clearer to me in terms of describing the aspects of entrainment, or the “dynamics of attending”.

  8. Several of you have commented on the apparent contrast between how entrainment is defined in these two articles. This is an important observation and we will return to this point tomorrow.

    Some questions have also been raised about how “meter” as opposed to beat is conceptualized, not only by Large and Jones (1999) in particular, but also as represented by experimental psychological research. In many studies, what passes for “metric” entrainment appears to be only a beat (or one pulse level), although when second- rather than first-order periodicity is involved (i.e., varying rhythmic patterns with some form of underlying regularity), one might argue that beat induction is the by-product of some form of interaction between at least two duration series (i.e., two partially realized pulses).

    I assigned the Large and Jones (1999) in part because it is a landmark, widely-cited article. However, Large has also written about his model elsewhere and I would call you attention to his 2000 article, “On synchronizing movements to music” (see classes*v2, Resources>Weekly Readings>Attention, Entrainment & SMS). In the first four pages, I have he summarizes the issue and explains how his model can explain meter perception. The most rewarding passage, from that perspective, is the last paragraph on p. 529 (page 3 of the PDF).

  9. Large 2000 is a good piece, and it does clear up some of my questions. That said, I remain uncertain as to whether Large intends to model an ‘autonomous’ system, that is, a perception, or perceptions, that occurs beneath the threshold of consciousness. Never does he explicitly say so, but then again, it may not occur to him that he must. On p.532, he writes, “Metrical structure, or simply meter, refers to the temporal pattern that is created by the simultaneous perception of beats” and so on. His using the passive voice here strikes my own sensitivities as especially interesting. Why not: “the simultaneous perception of beats creates metrical structure.” Well, I am not sure, but suggest that when the verb agent is revealed the idea does not ring truly. I might prefer tentatively: “the simultaneous perception of beats provides the raw material which IF processed conceptually/consciously can constitute a meter (or a few/several/many meters)”.

    – S P G

    • Why does it matter whether “meter” is unconscious or conscious? What is left out when the focus in on the unconscious processes? What are the questions that are not satisfactorily “answered” by his model?

      I would say that Large, here, is most likely focused on the unconscious processes by which these different beats manifest in the brain. But I am also pretty sure that he does not mean to negate the existence of conscious (or conceptual) meter. It has been shown in several different studies that a listener can impose a metrical structure on auditory stimuli and cause the brain to respond accordingly. The beat tracking experiment I conducted using Carter’s 90+ also suggest that musicians trained in 20th-century Western European art music can successfully impose such a framework even on highly irregular rhythmic sequences (that other listeners without this training might perceive as random).

      One might wonder, though, whether the conceptual could be enacted without the mechanisms Large attempts to model…

Comments are closed.