OMS Formal Exposition

OMS - Formal Exposition
12/2009 [under revision]

Table of Contents:

I. Overview
        A. Introduction
        B. Definition of OMS
        C. OMS is a model with 6 levels
        D. Terminology

II. The 6 Levels of OMS, in Further Detail
        A. Level 1 "Vocabulary"
        B. Level 2 "Intelligence" (The Society of Mind model of human intelligence)
        C. Level 3 "How-To"
        D. Level 4 "Customization"
        E. Level 5 "Foundation"
        F. Level 6 "Meta"

I. Overview

A. Introduction

The following is a formal academic-style presentation of the subject of OMS. This article assumes that the reader has already reviewed the introductory materials and some of the sample analyses on the website www.omsmodel.com

B. Definition of OMS

OMS is a model of the phenomenon of music. It can be applied to any work of music. A "work of music" is defined to be an individual musical performance. (OMS certainly sheds light on the matter of musical compositions/scores as well; but strictly speaking, OMS is concerned with the individual performance.)

OMS treats a work of music as follows:

[a] A work of music is regarded as an artifact - i.e., it is made by a human being.

[b] The work is treated abstractly as a set of stimuli that operate on a listener.

[c] OMS assumes a set of musical receptors ("receptors") which respond to the stimuli. Click here to see a Master List of important musical receptors . The concept of "musical receptor" (also sometimes called "agent", "resource", "processor") is discussed further in II-B below.

[d] OMS assumes that a work of music includes a large set of stimuli, and that these stimuli are highly organized. Stimuli are assumed to act upon a listener simultaneously and in a coordinated fashion. ("OMS" stands for "Organized Multi-Stimulation")

I.e. a work of music is modeled as an artifact which provides organized multi-stimulation of musical receptors. (This is the OMS "definition" of music.)

OMS may be applied to entire musical works or portions thereof.

Comments on the above:

[a] OMS is not literally a definition of music. Strictly speaking, music would have to defined as a kind of sound, not as a set of stimuli.

[b] OMS is a model (A model is not the same thing as a definition or a theory). It is outside the scope of this article to discuss the subject of models in general. But, the intention of OMS as a model of music is as follows:

[i] It provides a systematic representation of a broad range of important characteristics of a work of music.

[ii] It is a useful tool for analyzing individual works of music.

[iii] It is a useful tool for the analysis of many general issues/topics about music.

[iv] It is transparent and modifiable/customizable (see section C below).

Being a model, OMS intentionally is a somewhat simplified/schematized representation of music. The tradeoff for this is clarity, transparency, and relatively simplicity. (Notwithstainding its simplifications, OMS seems to be a very rich and wide-ranging model.)

C. OMS is a model with 6 levels

The OMS model is organized in 6 levels, the top level (Level 1) being the easiest to use, and the lower levels being increasingly profound and difficult.

The following is a summary description of the 6 levels. These levels will be discussed in greater detail in later sections of this article.

Level 1: "Vocabulary"

Level1 provides a Master List of about 90 musical terms or "tags" (some people would call them "attributes") that can be applied to any musical performance. The tags are straightforward to understand and easy to apply. The resulting profile provides a broad and deep understanding of the performance. Click here for an annotated list of these tags.

The tags are concerned not with notes or note-structures, but rather with how a performance acts upon a listener.

The list was constructed based on deep theory combined with analysis of many musical masterpieces, classical and popular. It provides a rich description of any kind of music. It is intended to encompass the major high-level categories in which a listener can experience music. Accordingly, we call this the Master List of tags. Depending on circumstances, a user may want to modify this list or extend it. But we call it a "Master List" to suggest that it is a good starting point for any other list of tags that a user may want to develop.

A user of the model could use the Master List of tags without being familiar with rest of OMS (Levels 2 and below). But a user will benefit greatly from knowledge of the rest of OMS.

Level 2: "Intelligence"

OMS includes a model of human intelligence called "Society of Mind". This model was developed by Marvin Minksy, the renowned pioneer in Artificial Intelligence ("AI"). Society Of Mind is well-known in the scientific community; it embodies decades of intensive R&D in the AI community; it is excellent for doing high-level models of human intelligence (this does not mean that SocietyOfMind models are the final word on human intelligence). Additionally, it turns out that certain materials in SocietyOfMind are excellent for the purpose of modeling music.

The primary reference for Minsky's model is his book The Society of Mind . The central idea of Society of Mind is that human intelligence consists of a large number of rather primitive processing mechanisms of wide variety, that are loosely coordinated/integrated. Although these individual mechanisms are mainly rather weak, in combination they can produce immense intelligence. In his original book, Minsky calls these mechanisms agents; more recently, Minksy calls them resources.

There are certain agents which are sensitive to music. For instance there is an agent which is sensitive to the range of loud/soft; another agent is sensitive to the range of high/low; another is sensitive to sonic richness; another is sensitive to beginnings; another is sensitive to endings; and so on. OMS holds that there is a Master List of about 90 high-level agents (these correspond to the tags in Level 1 above). OMS characterizes music (a musical performance) as something that stimulates musical agents in an intense, highly-coordinated manner. ("OMS" stands for "Organized Multi-Stimulation").

SocietyOfMind provides guidance for: Identifying agents (There are many more musical agents than the ones referenced in the Master List, the Master List is just the most important musical agents); thinking about how agents respond to stimulation; thinking about how agents affect each other; thinking about the "Great Questions of Music" (questions about meaning, emotion, greatness, authenticity, ...); and more.

SocietyOfMind will be discussed further below in section II-B.

Level 3: "How-To"

This level provides guidelines and suggestions about how to utilize Level1 and Level2. For instance: How to identify additional agents and how to relate them to the Master List; constructing a complete analysis of a musical performance; dealing with issues of objectivity/subjectivity; relating a composition/score to a performance thereof (OMS deals primarily with performances); evaluating music via comparisons or via composite numeric scores.

Level 4: "Customization"

OMS is a model, not a monolithic theory. It is designed to allow customization. This level outlines possible ways that Level1, Level2, Level3 might be customized.

Level 5: "Foundation"

OMS was developed from a methodical analysis of great music, combined with materials from classic academic philosophy and AI. Because of this, OMS analyses are almost always cogent and interesting - even if one disagrees with the OMS analysis, still the OMS carries the value of significant foundational materials.

In Level5, we outline the foundational materials that underlie OMS.

Level 6: "Meta"

Although OMS is rich and broad-ranging, it does not try to address the entirety of the phenomenon of music. This level provides critiques of OMS; it outlines areas that are not (fully) addressed by OMS; it points to possible models or theories that are outside the scope of OMS or somewhat opposed to OMS.

D. Terminology

Depending on context, we use the following terms more or less interchangeably: "agent", "resource", "receptor", "processor", "processing center". The terms "agent" and "resource" are used by Minsky re SocietyOfMind. We frequently use the term "receptor" to emphasize that most musical agents are primarily passive in nature.

Since every agent/receptor has a name and a function, each agent/receptor also amounts to a piece of terminology or vocabulary that can be used in describing music. In this context, we sometimes speak of "vocabulary", "attributes", or "tags".

II. The 6 Levels of OMS, In Further Detail

A. Level 1: "Vocabulary"

As indicated above, each tag in the Master List corresponds to an agent (in the sense of SocietyOfMind). E.g., there is a tag in the Master List for "beauty". So we would say that there is an agent which responds to beauty in music (when there is beauty in a piece of music, it generally stimulates this agent).

B. Level 2: "Intelligence" (Society of Mind model of human intelligence)

[1] The concept of an agent

[a] The concept of an agent has evolved out of efforts in AI to construct models of human intelligence that can be embodied in software. Minksy and other practitioners have concluded that human intelligence is best modeled as a large number of mildly-intelligent specialized processes that are loosely integrated.

[b] Some basic statements about agents:

[i] Any agent has a specific function which is easily described.

[ii] Agents can operate in combination with other agents.

[iii] Some kinds of agents may transmit information/signals to other agents; or vice versa.

[iv] Some agents can control/affect other agents; or vice versa.

[v] Agents can be discussed using psychological principles. E.g., agents can learn; agents can stimulate other agents; agents can be developed (or perhaps even created) via appropriate stimuli.

[vi] In general, agents do not correspond to a specific physical component of the brain.

[vii] Most high-level intelligent activity is the result of a large number of agents operating in parallel, and loosely coordinated.

[viii] Informal terminology that is sometimes used in place of "agent": receptor, resource, mental-resource, processor. If an agent is mainly passive, it could be called a "receptor".

[ix] In different human beings, the same agent may respond differently to the identical stimulus. In the same human being, a single agent may respond differently to an identical stimulus at different times.

[c] The concept of an agent has achieved acceptance for at least two reasons. First, it has proven useful in modeling human intelligence for the purpose of developing AI software. Second, the concept has achieved significant acceptance in the psychological community.

[2] Musical agents

[a] Most of the agents in the Master List are rather primitive in function. So, for instance if the contrapuntal receptor has a strong reaction to a piece of music, this does not necessarily imply that there is strong contrapuntal content in the classic music-theoretic sense (the receptor could be somewhat deceived by music that seemingly has more contrapuntal content than is really present). Or there could be a piece of music with significant contrapuntal content (in the class music-theoretic sense) which does not elicit a strong reaction from the contrapuntal receptor - perhaps because the contrapuntal materials are not easily observed or "felt" by a specific listener .

[b] When a musical agent is stimulated, sometimes this results in observable behavior by the listener. Examples of observable behavior: Changes in facial expression; moving hands, tapping feet, etc; dancing; applause at the end of a piece; talking about the piece after the performance. Sometimes stimulation may not immediately result in easily observed behavior, but may stimulate a disposition to certain behavior (e.g., a piece stimulates a disposition to talk about the piece, if the listener is prompted after the performance).

[c] OMS also takes the position that stimulation may result in rich subjective experience which may be difficult or impossible for a third party to observe. OMS takes this position in order to do justice to the strongly held belief of many listeners that music elicits a rich inner experience which is far beyond what could be described by a third party or even by the listener ("ineffable experience"). (We leave it an open question whether eventually all such subjective experience could be described/elucidated by scientific means.)

[d] Since musical receptors are primary passive (they receive a lot of input but they do not do a lot), we often call them receptors .

[3] What agents are there? What musical agents are there?

Minksy's view seems to be that the existence of agents is dependent on the purpose of the model. I.e., if one is using SocietyOfMind to develop a high-level architecture for general purpose robotics, then there will be a need for high-level agents such as find, get, put and so on. And then these high-level agents may be related to lower-level agents, and so on.

For the purpose of modeling music, we believe that the following are some of the principles for defining agents:

[a] Some aspects of sound-stimulation are musically significant, some are not. For instance, if a piece causes reactions of beauty, spatiality, emotion , these are generally considered musically significant. If a piece stimulates a listener's thyroid gland, this by itself would not generally be considered musically significant.

[b] The concept "musically significant" is not a simple to analyze. It would appear to be a value-concept rather than a scientific concept. The concept seems to evolve somewhat over time; it appears that one factor in its evolution is the influence of the best musicians.

[c] There does seem to be significant consensus regarding what kinds of musical stimulation are musically significant. This consensus can be documented to a degree by: [i] Reviewing the writings of the best-regarded musical critics and analysts; [ii] analyzing performances by highly-regarded musicians, analyzing scores of highly-regarded composers.

[d] There are some psychological or philosophical principles which probably bear on the selection of musical agents. E.g. "Humans naturally desire a degree of clarity in music that they listen to. Accordingly, there is a sensitivity to beginning; ending; signposts or markers within the piece."

Based on [c] and [d], the Master List of approximately 90 musical tags (agents) was developed (via a lengthy research project). There are many more musical agents than 90; but it appears that most other (candidate) agents can be related to ones on the Master List; or they are agents that are stimulated infrequently (e.g. the humor agent - there are few musical pieces or portions thereof that are genuinely humorous).

Using the above principles [a] - [d] as a guide, we speculate that 1000 - 3000 musical agents could be named and described in a straightforward way. (This is in addition to the 90 agents in the Master List). [[a few examples]]

However, for the purposes of standard OMS model, we focus on the Master List. And then users are free to revise/extend the Master List as appropriate.

C. Level 3: "How-To"

The following are default procedures. A user may want to revise these procedures as appropriate. Before revising a procedure, it is recommended that the user first get acquainted with the entire OMS model, and try the default procedure. This will give the user a better sense for the pros and cons of possible revisions.

[1] How to construct an OMS Profile of a musical performance

Construct a grid corresponding to the Master List (90 rows total). For each agent/receptor, assign a number from 0-4 based on the following:

0: Negligible stimulation
1: Noticeable stimulation
2: Strong stimulation
3: Creative stimulation: Rich, containing novel or highly-structured elements; strongly stimulates other receptors
4: Superlative; almost in a class by itself

The Master List includes provision for up to 5 "Wild Card" agents/receptors. Use these for agents which are not in the Master List, but which seem important for the piece you are profiling.

We are not currently providing formal guidance or instruction how to assign these numbers (although we think this could be done). Possible ways to assign the numbers: [a] Base the assignment on your own personal instantaneous reaction to a performance; [b] base the assignment on a more reflective reaction to the performance (try to factor out issues such as wandering attention, personal dislikes, ...); [c] try to take a step back and evaluate based on what the music did or tried-to-do to you, as opposed to how you reacted.

One way to enhance to value of OMS profiles is to profile several pieces of music side-by-side. Or do your own profiles of pieces profiled on this website, and then compare.

For examples, see Table of Contents, Section B .

[2] How to calculate a summary numeric score for a profile

For each of the 90 rows in the profile, calculate the square of the score (e.g. 0^2=0, 1^2=1, 2^2=4, 3^2=9, 4^2=16); then add these squares together.

Rough interpretation of this summary numeric score: It is a rough gauge of the total stimulation provided by the performance. Anything over 100 is usually very stimulating, and it is pleasurable in many respects. Most popular music is in the 100 - 200 range. Good classical music starts at about 200; great classical music starts at about 300; outstanding classical music (an unforgettable experience) starts at about 400. The highest rating found so far is for an outstanding performance of Beethoven Piano Sonata Op. 110, score of 680.

The above is "experimental", we do not currently have an exact interpretation to offer.

Many examples are in Table of Contents, Section B , also see the 30+ examples in OMS Music Blog .

[3] Suggestion for developing a complete analysis of a performance, based on OMS

For examples, see Table of Contents, Section B .

[a] Create a Profile, as described above

[b] Write a brief essay to accompany the Profile. Possible materials for an essay: Types of stimulation that are of unusual interest; highly primitive reactions that may be stimulated by the music; ways in which the music provokes "deep thought" (thoughts about time, space, existence, spirituality, ...).

Regarding "deep thought": It appears that most music - even simple popular music or childrens music - naturally stimulates a certain amount of deep thought. The deep thoughts stimulated would tend to be primitive in nature, but they are "deep". See the Master List of receptors/tags, especially sections 1,2 . In the author's personal opinion, any thorough analysis should take into account some of the deep thoughts stimulated by the music.

[4] Using OMS to analyze the Great Questions of Music

Click here to see some examples . OMS is useful for analysis of many of the Great Questions, because it often provides "answers" that are clear, transparent, succinct, and well-founded. They are well-founded because OMS itself has excellent foundations (see II-E). So even if one disagrees with an OMS "answer", the answer can be helpful for developing one's own views.

D. Level 4: "Customization"

Possible ways in which OMS can be customized:

[1] Revise or add to the Master List of tags/receptors

See B3 above for suggestions on how to do this.

[2] Revise the scoring system for constructing a Profile or calculating a summary numeric score for a piece.

[3] Elaborate the model for how an agent/receptor operates

In some cases, it might be helpful to model an agent as being able to respond in a multidirectional manner - e.g. the agent for beauty might be characterized to respond positively to beauty and negatively for ugliness.

With some agents, it may be helpful to split the agent into a family of 2 or more agents. E.g. the Genesis Effect agent might be split into a family of agents that relate to different kinds of Genesis Effects ("male Genesis Effect"; "female Genesis Effect"; ...)

E. Level 5: "Foundations"

The following is an outline of foundational considerations in support of OMS:

[1] Late Wittgenstein

In his book Philosophical Investigations and other work from his later period, the philosopher Ludwig Wittgenstein demonstrated a technique for investigating deep/complex concepts (such as knowledge, certainty, meaning, existence, understanding). Although there are many interpretations of Wittgenstein work, the following position seems central to Wittgenstein's work:

To understand a deep/complex concept, it is often helpful to observe how the concept operates in ordinary life. Specifically, it is useful to observe: Situations in which the concept is employed; the "profile" of each such situation (social, behavioral, circumstantial, …); how the concept is functions in actual language, and how this language functions in actual situations.

I will call this the Anthropological View (AV) of concepts. A straightforward way to apply AV to music would be as follows:

To understand the concept music, observe how music functions in ordinary life. Doing this, the following aspects are clear: Music is an artifact; music ordinarily involves sound; music stimulates a listener in many different ways. Some kinds of stimulation are known only by verbal reports from the listener. However, many kinds of stimulation are observable, e.g.: The listener moves physically; the listener may verbalize; the listener may make other sounds; after a performance, it is normal for a listener to applaud or respond in some alternate way; after a performance, it is normal for a listener to want to talk about the performance.

The OMS model is closely related to the above view of music.

[2] Artificial Intelligence and SocietyOfMind

See II-B above.

[3] Anti-reductionism

"Reductionism" is roughly defined as any attempt to explain a complex phenomenon in terms of a small number of basic concepts. Examples of reductionism:

[i] Utilitarianism: All correct principles of ethics may be derived from a single principle: the principle of "greatest total utility" for the relevant population.

[ii] Axiomatization of arithmetic: All truths about arithmetic may be derived from a small number of axioms.

[iii] Universal notation for all pronunciation: The pronunciation of any word in any recognized language may be specified using the International Phonetic Alphabet.

[iv] Subatomic physics: All matter is composed from a small number of kinds of subatomic particles.

[v] Note-centric model of music: All significant aspects of music may be modeled in terms of the concept of a note plus concepts that can be derived from it (e.g. motive, phrase, melody, chord, rhythm, tone center, Schenker structures, …)

Reductionism has been popular in Western thought for many centuries; but its popularity has diminished in recent decades. For instance, it is now known that [ii] is simply false, without any plausible workaround; [i] is largely discredited. There are of course some examples of fairly successful reductionism, e.g. [iii], [iv] .

How about [v] the note-centric model of music? The modern history of western music has been virtually monopolized by the note-centric model, and there is no question that the note-centric model is highly productive. However, in view of recent anti-reductionist tendencies in Western thought, it is natural to examine the note-centric model and to look for alternatives that are not reductionist. In this context, OMS is a natural, even "mainstream" anti-reductionist way of modeling music.

(It is outside the scope of this article to discuss the history of reductionism in depth. Also it is outside the scope of this article to provide a detailed critique of the note-centric model of music. However, critiques of the note-centric model have been expressed elsewhere.)

[4] Actual musical discourse

Most of the items in the OMS Master List can be found in active use in ordinary musical discourse. It could be said the ordinary musical discourse frequently resembles portions of OMS - but applied in a non-systematic fashion. I.e., one of the contributions of OMS is to provide a proper organizational scheme and foundation for things that are already occurring in ordinary musical discourse. One might view OMS as a highly-enlightened, highly-organized, highly-articulate "musical critic".

[5] Actual musical practice

Although we won't attempt to "prove" this, it seems to this author that the idea of organized multi-stimulation of musical agents is often implicit in the work of performers and composers. Many of the agents in the Master List were first found in analysis of Beethoven symphonies. A great deal of professionally produced popular music is quite overt about its aim of stimulating listeners methodically in specific ways.

[6] Additional considerations from classic academic philosophy

[[additional materials, tbd]]

F. Level 6: "Meta"

[[This section is a preliminary sketch, still under development.]]

Although OMS is rich and broad-ranging, it does not address the entirety of the phenomenon of music. This level provides critiques of OMS; it suggests areas that may not be (fully) addressed by OMS; it points to possible models or theories that are outside the scope of OMS or somewhat opposed to OMS.

[1] OMS may not fully address the musical composition/score

Possible View#1: A musical score is addressed to prospective performers; the score is a performance-specification that is more or less detailed in various respects. If this view is accepted, then OMS could conceivably provide a fairly good analysis of many musical scores.

Possible View#2: A musical score is (sometimes) something altogether different than View#1. Perhaps it could be characterized as the crystallization of a composer's intense creative process; and it relates most closely to something "inside the composer" as opposed to a performance-specification for a performer. If View#2 is accepted, then OMS may not provide a good analysis of the musical score.

[2] OMS is different from the "Note-centric" model of music

The "Note-Centric" model of music is that: "All significant aspects of music may be modeled in terms of the concept of a note plus concepts that can be derived from it (e.g. motive, phrase, melody, chord, rhythm, tone center, Schenker structures, ...)" . This is of course the dominant method for modeling Western music and a great deal of non-Western music.

It does not appear that OMS is in conflict with the Note-Centric model. Probably the Note-Centric model and OMS are simply two different models for a huge and complex subject-matter. (Minksy's view is that for complex subject-matter, multiple models are appropriate).

[3] Is music stimulation subject to saturation, or is it open-ended?

This is a difficult issue even to formulate. Here is a rough formulation: If we were to accept the metric described in II-C-2, then the maximum stimulation possible for any piece of music would be about 1000 (the highest identified so far is 680). But does it make sense that there is such a thing as maximum possible stimulation with regard to music? Perhaps we need an alternative way to characterize musical stimulation so that the possible stimulation is (virtually) unlimited! Perhaps OMS goes too far in characterizing music-listening as a primitive, "sensual" activity, and it does not do justice to music-listening as a high-level (intellectual, spiritual) activity.

[4] With musical stimulation, is "quality" sometimes more important than "quality"?

[5] Is it possible that a great deal of musical stimulation is at a "micro-level" than cannot be characterized by SocietyOfMind/agents?

[6] Objectivity vs Subjectivity in music analysis

There are alternatives to how OMS deals with issues of objectivity vs subjectivity in music analysis. (Music theory is permeated with issues about objectivity and subjectivity).