pith. sign in

arxiv: 2604.07476 · v1 · submitted 2026-04-08 · ❄️ cond-mat.stat-mech

Rhythm as an ordered phase of sound: how musical meter emerges in a statistical mechanical model

Robert St.Clair , Jesse Berezovsky This is my paper

Pith reviewed 2026-05-10 17:00 UTC · model grok-4.3

classification ❄️ cond-mat.stat-mech
keywords musical rhythmmeterphase transitionsstatistical mechanicsmean field approximationeffective free energyBach compositionsrhythm generation
0
0 comments X

The pith

Musical meter emerges as an ordered phase when a model balances human preferences for repetition against variety in an effective free energy.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a statistical mechanics model that treats the human preference for repeated rhythmic patterns as an energy term and the desire for complexity as an entropy term. These are combined into an effective free energy that is minimized in the grand canonical ensemble. A mean-field approximation then produces phase transitions from disordered timing to ordered structures that resemble musical meter. The resulting rhythmic statistics show quantitative agreement with a dataset of Bach compositions. This framing suggests rhythm can be derived from a physical optimization principle rather than imposed solely by convention.

Core claim

By mapping competing psychological preferences onto energy and entropy contributions, the model defines an effective free energy minimized in the grand canonical ensemble. Mean-field calculations of this free energy yield phase transitions from temporally disordered events to ordered rhythmic patterns. These ordered states reproduce key characteristics of musical meter, and their statistical properties align well with those observed in compositions by Bach.

What carries the argument

Effective free energy in the grand canonical ensemble, with repetition as an energy term and variety as an entropy term, whose mean-field minimization produces phase transitions to rhythmic order.

If this is right

  • The ordered phases correspond to common meters and rhythmic features found in actual music.
  • Rhythmic order can be generated directly from the minimization without additional cultural rules.
  • Small shifts in the balance between repetition and variety can trigger abrupt changes in rhythmic structure.
  • The model supplies a quantitative method for producing new rhythms consistent with observed musical patterns.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same free-energy construction might be applied to model rhythmic preferences in non-Western traditions or historical style changes by adjusting the relative weights.
  • Direct psychological experiments that measure perceived naturalness for different degrees of repetition could test whether human judgments track the model's phase boundaries.
  • The approach could extend to other ordered temporal sequences, such as metrical patterns in poetry or periodic movement in dance.

Load-bearing premise

The specific energy and entropy terms chosen for the effective free energy accurately capture human psychological preferences for repetition versus variety without further empirical calibration.

What would settle it

A large corpus of listener preference data or additional musical datasets whose rhythmic statistics fall outside the model's predicted ranges for any choice of the free-energy parameters.

Figures

Figures reproduced from arXiv: 2604.07476 by Jesse Berezovsky, Robert St.Clair.

Figure 1
Figure 1. Figure 1: FIG. 1: Illustration of hierarchical meter found in music. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2: The two-site ( [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3: (a)-(e) show results of the eight-site ( [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figure 1
Figure 1. Figure 1: With all probabilities significantly deviating from zero or one, a wider range of note [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4: Phase diagram of 8-site ( [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5: Probabilities [PITH_FULL_IMAGE:figures/full_fig_p015_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: As µ becomes more negative, the regular repetition of the ground state is main￾tained, but with the periodicity changing. Lower µ favors an equilibrium state with a lower concentration of events. The ground state in the black region corresponds to all pl → 1, whereas the ground state in the blue region corresponds to every 8th pl → 1, and all other pl → 0. The wedge-shaped regions of bistability seen at lo… view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6: Probabilities [PITH_FULL_IMAGE:figures/full_fig_p017_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7: Probabilities [PITH_FULL_IMAGE:figures/full_fig_p021_7.png] view at source ↗
read the original abstract

We develop a model of musical rhythm and meter based on optimizing the trade-off between human psychological preferences for perceiving repeated patterns in time with a desire for variety and complexity. By mapping these competing preferences onto analogous quantities in statistical physics, we define an effective free energy which is minimized in the grand canonical ensemble. Using a mean field approximation, we observe phase transitions in the model from disordered events in time to orderings that closely reproduce those seen in music. We then compare the range of rhythmic characteristics predicted by the model to a dataset drawn from compositions by Johann Sebastian Bach, finding generally good quantitative agreement. The results provide a new lens through which to study musical rhythm, and a method for generatively producing rhythms.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 0 minor

Summary. The manuscript develops a statistical mechanical model of musical rhythm and meter by mapping human psychological preferences for repetition versus variety onto an effective free energy functional, which is then minimized in the grand canonical ensemble. A mean-field approximation is used to identify phase transitions from disordered temporal events to ordered states that reproduce musical meter patterns, followed by a quantitative comparison to rhythmic features extracted from compositions by J. S. Bach, with the claim of generally good agreement.

Significance. If the mapping from psychological preferences to the free-energy terms can be independently validated, the work offers a fresh physics-based perspective on the emergence of musical order as a phase transition, with potential implications for generative rhythm models and interdisciplinary studies of perception. The approach is conceptually innovative in treating meter as an ordered phase, but its significance is currently limited by the absence of explicit model equations and validation details that would allow assessment of robustness.

major comments (3)
  1. [Abstract and model section] Abstract and model definition: the central claims of phase transitions and quantitative agreement with Bach data are asserted without providing the explicit form of the effective free energy, the mean-field self-consistency equations, the fitting procedure for any parameters, or the precise metrics used for the data comparison. This prevents verification of whether the observed orderings follow from the minimization or are imposed by construction.
  2. [Model definition] The effective free energy is constructed directly from the repetition-favoring energy term and variety-favoring entropy term that encode the target psychological preferences; any emergent ordered phases are therefore shaped by the same functional choices the model seeks to explain. This creates a circularity that must be addressed by showing the phase diagram is robust to reasonable variations in the functional forms or by providing independent psychological calibration data.
  3. [Comparison to Bach data] The manuscript reports a single free parameter yet claims generally good quantitative agreement with Bach rhythms; without the explicit comparison procedure or sensitivity analysis (e.g., how the agreement metric changes when the repetition-variety trade-off is varied), it is unclear whether the match is a genuine prediction or an artifact of parameter choice.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their thoughtful review and constructive suggestions. We appreciate the recognition of the conceptual innovation in treating meter as an ordered phase. Below, we address each major comment point by point, indicating the revisions we will make to the manuscript.

read point-by-point responses

  1. Referee: [Abstract and model section] Abstract and model definition: the central claims of phase transitions and quantitative agreement with Bach data are asserted without providing the explicit form of the effective free energy, the mean-field self-consistency equations, the fitting procedure for any parameters, or the precise metrics used for the data comparison. This prevents verification of whether the observed orderings follow from the minimization or are imposed by construction.

    Authors: We agree that the manuscript would benefit from greater explicitness in these areas. In the revised version, we will expand the model section to include the explicit mathematical form of the effective free energy functional, derive the mean-field self-consistency equations in detail, describe the procedure for fitting the single free parameter, and specify the quantitative metrics (such as pattern frequency distributions and correlation measures) used for comparing model predictions to the Bach dataset. This will allow readers to verify the emergence of the ordered phases directly from the minimization. revision: yes

  2. Referee: [Model definition] The effective free energy is constructed directly from the repetition-favoring energy term and variety-favoring entropy term that encode the target psychological preferences; any emergent ordered phases are therefore shaped by the same functional choices the model seeks to explain. This creates a circularity that must be addressed by showing the phase diagram is robust to reasonable variations in the functional forms or by providing independent psychological calibration data.

    Authors: We acknowledge the concern regarding potential circularity in the model construction. The energy and entropy terms are indeed motivated by psychological preferences documented in the literature on music perception. To mitigate this, we will add an analysis demonstrating the robustness of the phase transitions and ordered states to reasonable variations in the functional forms of the repetition and variety terms. While independent psychological calibration data for the specific parameters is not currently available in our study, the robustness check will help establish that the emergent meter patterns are not solely an artifact of the chosen forms. revision: partial

  3. Referee: [Comparison to Bach data] The manuscript reports a single free parameter yet claims generally good quantitative agreement with Bach rhythms; without the explicit comparison procedure or sensitivity analysis (e.g., how the agreement metric changes when the repetition-variety trade-off is varied), it is unclear whether the match is a genuine prediction or an artifact of parameter choice.

    Authors: We will revise the comparison section to provide a detailed description of the procedure used to compare the model's rhythmic characteristics with the Bach compositions, including how the single free parameter is selected (e.g., via optimization against a subset of the data or based on prior literature). Additionally, we will include a sensitivity analysis showing how the agreement metrics vary with changes in the repetition-variety trade-off parameter, thereby clarifying that the agreement reflects genuine predictive power rather than parameter tuning alone. revision: yes

Circularity Check

1 steps flagged

Effective free energy defined by mapping target psychological preferences, so observed musical orderings emerge by construction from the input ansatz.

specific steps
  1. self definitional [Abstract]
    "We develop a model of musical rhythm and meter based on optimizing the trade-off between human psychological preferences for perceiving repeated patterns in time with a desire for variety and complexity. By mapping these competing preferences onto analogous quantities in statistical physics, we define an effective free energy which is minimized in the grand canonical ensemble. Using a mean field approximation, we observe phase transitions in the model from disordered events in time to orderings that closely reproduce those seen in music."

    The effective free energy is defined by directly mapping the psychological preferences (repetition-favoring energy term, variety-favoring entropy term) that the model seeks to explain. The subsequent minimization and phase transitions therefore produce orderings shaped by those same preferences by construction; the match to musical meter is not an independent derivation but a re-expression of the input mapping.

full rationale

The derivation begins by explicitly constructing the effective free energy from the very human preferences for repetition versus variety that the model is intended to explain. Minimization then yields phase transitions whose rhythmic characteristics are reported to match music. Because the functional forms encode the target behavior directly and no independent psychological calibration or external validation is invoked, the quantitative agreement with Bach data reduces to the initial mapping rather than an independent prediction. This is partial circularity (score 6) rather than total, as the mean-field calculation itself is non-trivial.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on an untested translation of psychological preferences into thermodynamic quantities and on the validity of the mean-field approximation for the resulting model.

free parameters (1)
  • repetition and variety trade-off parameters
    The relative weights of the repetition-rewarding and variety-penalizing terms in the effective free energy must be chosen or fitted; these choices directly control where the phase transition occurs.
axioms (2)
  • domain assumption Human preferences for repeated temporal patterns versus variety can be represented by additive energy and entropy contributions in an effective free energy.
    This mapping is invoked to define the model but is not derived from psychological data or first principles within the abstract.
  • domain assumption A mean-field approximation is sufficient to locate the phase transition between disordered and ordered rhythmic states.
    The abstract states that the mean-field calculation reveals the ordering; fluctuations beyond mean field are not discussed.

pith-pipeline@v0.9.0 · 5417 in / 1534 out tokens · 65112 ms · 2026-05-10T17:00:30.147417+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages

  1. [1]

    Over a grid of values spanning ranges inµandT, we calculatep l using the iterative method described above, starting from uniform random initial conditions

    Phase diagram for the 8-site model We explore the results of the 8-site model by constructing a phase diagram as a function ofµandT. Over a grid of values spanning ranges inµandT, we calculatep l using the iterative method described above, starting from uniform random initial conditions. The resultingp l are found to be symmetric by translation of 1, 2, 4...

    work page

  2. [2]

    common time

    Self-similarity By construction, the total rhythmicity per site ˜Rtot = 1 for any regularly repeating se- quence of events. Such a regularly repeating pattern is the ground state of this model, approached asT→0. This ground state appears at the left side of the phase diagram in Fig. 4. Asµbecomes more negative, the regular repetition of the ground state i...

    work page doi:10.5281/zenodo.19473679

  3. [3]

    W. E. Benjamin, Music perception1, 355 (1984)

    work page 1984

  4. [4]

    H. C. Longuet-Higgins and C. S. Lee, Perception11, 115 (1982)

    work page 1982

  5. [5]

    Lerdahl and R

    F. Lerdahl and R. S. Jackendoff,A Generative Theory of Tonal Music, reissue, with a new preface(MIT press, 1996)

    work page 1996

  6. [6]

    London,Hearing in time: Psychological aspects of musical meter(Oxford University Press, 2012)

    J. London,Hearing in time: Psychological aspects of musical meter(Oxford University Press, 2012)

    work page 2012

  7. [7]

    W. A. Sethares,Rhythm and transforms(2007) pp. 1–336

    work page 2007

  8. [8]

    M. J. Steedman, Perception6, 555 (1977)

    work page 1977

  9. [9]

    Povel and P

    D.-J. Povel and P. Essens, Music Perception: An Interdisciplinary Journal2, 411 (1985)

    work page 1985

  10. [10]

    Desain and H

    P. Desain and H. Honing, Computer Music Journal13, 56 (1989)

    work page 1989

  11. [11]

    Desain, Music Perception: An Interdisciplinary Journal9, 439 (1992)

    P. Desain, Music Perception: An Interdisciplinary Journal9, 439 (1992)

    work page 1992

  12. [12]

    E. W. Large and J. F. Kolen, Connection Science6, 177 (1994)

    work page 1994

  13. [13]

    Ravignani, B

    A. Ravignani, B. Thompson, M. Lumaca, and M. Grube, Frontiers in Computational Neuro- science12(2018)

    work page 2018

  14. [14]

    Handel and G

    S. Handel and G. R. Lawson, Perception & Psychophysics34, 103 (1983)

    work page 1983

  15. [15]

    M. L. Serafine, N. Glassman, and C. Overbeeke, Music Perception6, 397 (1989)

    work page 1989

  16. [16]

    Desain and H

    P. Desain and H. Honing, Perception32, 341 (2003)

    work page 2003

  17. [17]

    P. E. Keller and D. K. Burnham, Music Perception22, 629 (2005)

    work page 2005

  18. [18]

    E. W. Large and C. Palmer, Cognitive science26, 1 (2002)

    work page 2002

  19. [19]

    Jacoby and J

    N. Jacoby and J. H. McDermott, Current Biology27, 359 (2017)

    work page 2017

  20. [20]

    Kaplan, J

    T. Kaplan, J. Cannon, L. Jamone, and M. Pearce, PLOS Computational Biology18, e1010579 (2022)

    work page 2022

  21. [21]

    Jacoby, R

    N. Jacoby, R. Polak, J. A. Grahn, D. J. Cameron, K. M. Lee, R. Godoy, E. A. Undurraga, T. Huanca, T. Thalwitzer, N. Doumbia, D. Goldberg, E. H. Margulis, P. C. Wong, L. Jure, 29 M. Rocamora, S. Fujii, P. E. Savage, J. Ajimi, R. Konno, S. Oishi, K. Jakubowski, A. Holzapfel, E. Mungan, E. Kaya, P. Rao, M. A. Rohit, S. Alladi, B. Tarr, M. Anglada-Tort, P. M....

    work page 2024

  22. [22]

    E. E. Hannon and S. P. Johnson, Cognitive Psychology50, 354 (2005)

    work page 2005

  23. [23]

    Berezovsky, Science Advances5, 10.1126/sciadv.aav8490 (2019)

    J. Berezovsky, Science Advances5, 10.1126/sciadv.aav8490 (2019)

    work page doi:10.1126/sciadv.aav8490 2019

  24. [24]

    Din and J

    H. Din and J. Berezovsky, PLoS ONE18, 10.1371/journal.pone.0280227 (2023)

    work page doi:10.1371/journal.pone.0280227 2023

  25. [25]

    Buechele and J

    R. Buechele and J. Berezovsky, Physical Review E110, 014145 (2024)

    work page 2024

  26. [26]

    Buechele, A

    R. Buechele, A. Cooke, and J. Berezovsky, Empirical Musicology Review19, 118 (2024)

    work page 2024

  27. [27]

    J. E. Youngblood, Journal of Music Theory2, 24 (1958)

    work page 1958

  28. [28]

    Knopoff and W

    L. Knopoff and W. Hutchinson, Journal of Music Theory27, 75 (1983)

    work page 1983

  29. [29]

    Hutchinson and L

    W. Hutchinson and L. Knopoff, Music Perception4, 281 (1987)

    work page 1987

  30. [30]

    I. D. Colley and R. T. Dean, PLoS ONE14, 10.1371/JOURNAL.PONE.0216088 (2019)

    work page doi:10.1371/journal.pone.0216088 2019

  31. [31]

    D. J. Levitin, P. Chordia, and V. Menon, Proceedings of the National Academy of Sciences of the United States of America109, 10.1073/pnas.1113828109 (2012)

    work page doi:10.1073/pnas.1113828109 2012

  32. [32]

    L. Liu, J. Wei, H. Zhang, J. Xin, and J. Huang, PLoS ONE8, 10.1371/journal.pone.0058710 (2013)

    work page doi:10.1371/journal.pone.0058710 2013

  33. [33]

    Palmer and C

    C. Palmer and C. L. Krumhansl, Journal of Experimental Psychology: Human Perception and Performance16, 728 (1990)

    work page 1990

  34. [34]

    Temperley,Music and probability(MIT Press, 2007) p

    D. Temperley,Music and probability(MIT Press, 2007) p. 244

    work page 2007

  35. [35]

    Temperley and D

    D. Temperley and D. Sleator, Computer Music Journal23, 10 (1999)

    work page 1999

  36. [36]

    Huron,Sweet anticipation: Music and the psychology of expectation(MIT press, 2008)

    D. Huron,Sweet anticipation: Music and the psychology of expectation(MIT press, 2008)

    work page 2008

  37. [37]

    L. H. Marckx and T. Saks, ProQuest Dissertations and Theses (1998)

    work page 1998

  38. [38]

    J. T. Harlow, ProQuest Dissertations and Theses (1993). 30

    work page 1993