Octave generalization and the identification of distorted melodies

Four familiar melodies were distorted in such a way that certain harmonic information was maintained. These melodies and their nondistorted versions were randomly presented to three groups of musically untrained subjects. There were, thus, undistorted melodies which were equal in all ways except the frequencies of the constituent tones, three conditions in which only the octave was maintained, and three other harmonic control conditions in which only the perfect fifth, fourth, and major third were maintained. The melodic sequences were presented to each group using a different tonal waveform for each group: sine wave, rectangular wave, and sawtooth wave. An analysis of variance revealed that the octave distortions were significantly more recognizable than the control conditions. Furthermore, waveform significantly influenced identification of melodies. It was proposed that octave generalization was a major factor in the identification of the distorted octave sequences.     

Octave generalization and tune recognition

An experiment was performed to investigate whether or not we are able to use octave generalization in recognizing tunes. The firat half of the tune “Yankee Doodle” wu chosen as the test sequence. This was universally recognized when played in any one of three octaves. However, when the sequence was played in identical fashion. except that each note was chosen randomly from one of the same three octaves, the percentage correct recognition was not significantly different from that obtained when the sequence was played as a series of clicks with the pitch information omitted but the rhythmic information retained. It is concluded that tune recognition takes place along a channel which is independent from that which gives rise to octave generalization. The relevance of this finding to the theory of music recognition proposed by Deutsch (1969) is discussed.     

Octave generalization of specific interference effects in memory for tonal pitch

An experiment is described which demonstrates that certain highly specific disruptive effects in immediate memory for tonal pitch generalize across octaves. Pitch recognition was required after a retention interval during which six other tones were played. The effects of including in the interpolated sequence tones which were removed by exactly an octave from those which had already been demonstrated to produce disruption were investigated. Generalization of these interference effects was found to result both from tones which were displaced an octave higher and (to a lesser extent) from tones which were displaced an octave lower. It is concluded that the memory store that retains information concerning the pitch of ifidividual tones is bidimensional in nature, both “tone height” and “tone chroma” being represented.     

Conceptual and perceptual information both influence melody identification

Three processes have been identified as central to object identification: top-down processing, bottom-up processing, and lateral competition. Six experiments using the perceptual interference paradigm were conducted to assess the relative contributions of these three processes to melody identification. Significant interference was observed only when the target and the distracting information were difficult to distinguish both perceptually and conceptually. Lateral competition-the activation of specific distractor melodies--did not influence the magnitude of interference observed. These results suggest that bottom-up and top-down processes contribute more to melody identification than does lateral competition. The data are discussed in terms of the broader literature on object identification and the relationship between identifying melodies, spoken words, and visual objects.     

Serial processing in melody identification and the organization of musical semantic memory

Unlike the visual stimuli used in most object identification experiments, melodies are organized temporally rather than spatially. Therefore, they may be particularly sensitive to manipulations of the order in which information is revealed. Two experiments examined whether the initial elements of a melody are differentially important for identification. Initial exposures to impoverished versions of a melody significantly decreased subsequent identification, especially when the early exposures did not include the initial notes of the melody. Analyses of the initial notes indicated that they are differentially important for melody identification because they help the listener detect the overall structure of the melody. Confusion errors tended to be songs that either were drawn from the same genre or shared similar phrasing. These data indicate that conceptual processing influences melody identification, that phrase-level information is used to organize melodies in semantic memory, and that phrase-level information is required to effectively search semantic memory.     

Octave equivalence as measured by similarity ratings

Octave equivalence refers to the musical equivalence of notes separated by one or more octaves; such notes are assigned the same name in the musical scale. The present series of experiments was an attempt to determine whether octave equivalence would be incorporated into subjects’ similarity ratings of pairs of tones or tone sequences. In the first experiment, subjects on each trial rated the similarity of two successively presented tones. The results failed to show evidence of octave equivalence. In subsequent experiments, the range of frequency values presented and the musical context were manipulated. Evidence of octave equivalence was found only when the range of tone height differences presented was small; the effect of musical context was negligible. The implications of these results for theories of music perception and recognition are discussed.     

Articulation effects in melody recognition memory

Various surface features—timbre, tempo, and pitch—influence melody recognition memory, but articulation format effects, if any, remain unknown. For the first time, these effects were examined. In Experiment 1, melodies that remained in the same, or appeared in a different but similar, articulation format from study to test were recognized better than were melodies that were presented in a distinct format at test. A similar articulation format adequately induced matching processes to enhance recognition. Experiment 2 revealed that melodies rated as perceptually dissimilar on the basis of the location of the articulation mismatch did not impair recognition performance, suggesting an important boundary condition for articulation format effects on memory recognition—the matching of the memory trace and recognition probe may depend more on the overall proportion, rather than the temporal location, of the mismatch. The present findings are discussed in terms of a global matching advantage hypothesis.     

Some structural determinants of melody recall

Sophisticated musicians were asked to recall, using musical notation, a set of unfamiliar folk tunes that varied in rhythmic structure and referents of tonality. The results showed that memory was facilitated by tonic triad members marking phrase endings, but only when their presence was highlighted by a corresponding pattern of temporal accents. Conversely, recall significantly declined when tonal information was either absent or obscured by rhythmic structure. Error analyses further revealed that the retention of overall pitch contour and information at phrase ending points varied as a function of these manipulations. The results are discussed in terms of a framework that links the acts of perceiving and remembering to a common attentional scheme.     

A probabilistic model of melody perception

This study presents a probabilistic model of melody perception, which infers the key of a melody and also judges the probability of the melody itself. The model uses Bayesian reasoning: For any "surface" pattern and underlying "structure," we can infer the structure maximizing P (structure|surface) based on knowledge of P (surface, structure). The probability of the surface can then be calculated as ∑ P (surface, structure), summed over all structures. In this case, the surface is a pattern of notes; the structure is a key. A generative model is proposed, based on three principles: (a) melodies tend to remain within a narrow pitch range; (b) note-to-note intervals within a melody tend to be small; and (c) notes tend to conform to a distribution (or key profile) that depends on the key. The model is tested in three ways. First, it is tested on its ability to identify the keys of a set of folksong melodies. Second, it is tested on a melodic expectation task in which it must judge the probability of different notes occurring given a prior context; these judgments are compared with perception data from a melodic expectation experiment. Finally, the model is tested on its ability to detect incorrect notes in melodies by assigning them lower probabilities than the original versions.     

Toward a technology of generalization: The identification of natural contingencies of reinforcement

Applied behavior analysts have directed a growing effort towards the development of a technology for behavioral generalization and maintenance over the past decade. Among the list of potential generalization promoters thought to exist is the natural contingency of social reinforcement (i.e., a behavioral trap) for new behavior in its untrained form or setting, or over time. Although past researchers have noted a need to program for the generalization and maintenance of behavior change, the current understanding of behavioral traps precludes the use of these contingencies to support behavioral changes when interventions are not in operation. This article describes five forms of evidence useful for the identification and analysis of natural contingencies of reinforcement. Examples from the applied research literature are provided to illustrate the kinds of studies that would greatly enhance our knowledge of behavioral traps and improve our ability to understand and program the generalization of trained behaviors across diverse forms and settings, and over time.     

Integration of melody and text in memory for songs

Two experiments examined whether the memory representation for songs consists of independent or integrated components (melody and text). Subjects heard a serial presentation of excerpts from largely unfamiliar folksongs, followed by a recognition test. The test required subjects to recognize songs, melodies, or texts and consisted of five types of items: (a) exact songs heard in the presentation; (b) new songs; (c) old tunes with new words; (d) new tunes with old words; and (e) old tunes with old words of a different song from the same presentation (‘mismatch songs’). Experiment 1 supported the integration hypothesis: Subjects' recognition of components was higher in exact songs (a) than in songs with familiar but mismatched components (e). Melody recognition, in particular, was near chance unless the original words were present. Experiment 2 showed that this integration of melody and text occurred also across different performance renditions of a song and that it could not be eliminated by voluntary attention to the melody.     

Conscious machines: Memory,melody and muscular imagination

A great deal of effort has been, and continues to be, devoted to developing consciousness artificially (A small selection of the many authors writing in this area includes: Cotterill (J Conscious Stud 2:290–311, 1995, 1998), Haikonen (2003), Aleksander and Dunmall (J Conscious Stud 10:7–18, 2003), Sloman (2004, 2005), Aleksander (2005), Holland and Knight (2006), and Chella and Manzotti (2007)), and yet a similar amount of effort has gone in to demonstrating the infeasibility of the whole enterprise (Most notably: Dreyfus (1972/1979, 1992, 1998), Searle (1980), Harnad (J Conscious Stud 10:67–75, 2003), and Sternberg (2007), but there are a great many others). My concern in this paper is to steer some navigable channel between the two positions, laying out the necessary pre-conditions for consciousness in an artificial system, and concentrating on what needs to hold for the system to perform as a human being or other phenomenally conscious agent in an intersubjectively-demanding social and moral environment. By adopting a thick notion of embodiment—one that is bound up with the concepts of the lived body and autopoiesis (Maturana and Varela 1980; Varela et al. 2003; and Ziemke 2003, 2007a, J Conscious Stud 14(7):167–179, 2007b)—I will argue that machine phenomenology is only possible within an embodied distributed system that possesses a richly affective musculature and a nervous system such that it can, through action and repetition, develop its tactile-kinaesthetic memory, individual kinaesthetic melodies pertaining to habitual practices, and an anticipatory enactive kinaesthetic imagination. Without these capacities the system would remain unconscious, unaware of itself embodied within a world. Finally, and following on from Damasio’s (1991, 1994, 1999, 2003) claims for the necessity of pre-reflective conscious, emotional, bodily responses for the development of an organism’s core and extended consciousness, I will argue that without these capacities any agent would be incapable of developing the sorts of somatic markers or saliency tags that enable affective reactions, and which are indispensable for effective decision-making and subsequent survival. My position, as presented here, remains agnostic about whether or not the creation of artificial consciousness is an attainable goal.     

Slow potentials in a melody recognition task

In a previous study, slow negative shifts were found in the EEG of subjects listening to well-known melodies. The two experiments reported here were designed to investigate the variables to which these slow potentials are related. In the first experiment, two opposite hypotheses were tested: The slow shifts might express subjects' acquaintance with the melodies or, on the contrary, the effort invested to identify them. To this end, some of the melodies were presented in the rhythms of other melodies to make recognition more difficult. Further, melodies rated as very well-known and as very unknown were analysed separately. However, the slow shifts were not affected by these experimental variations. Therefore in the second experiment, on the one hand the purely physical parameters intensity and duration were varied, but this variation had no impact on the slow shifts either. On the other hand, recognition was made more difficult by monotonously repeating the pitch of the 4th tone for the rest of some melodies. The slow negative shifts were enhanced with these monotonous melodies. This enhancement supports the "effort" hypothesis. Accordingly, the ofter shifts obtained in both experiments might likewise reflect effort. But since the task was not demanding, it is suggested that these constant shifts reflect the effort invested for coping with the entire underarousing situation rather than with the task. Frequently, slow eye movements occurred in the same time range as the slow potentials, resulting in EOG potentials spreading to the EEG recording sites. Yet results did not change substantially when the EEG recordings were corrected for the influence of EOG potentials.     

Expectancy in melody: tests of children and adults

Melodic expectancies among children and adults were examined. In Experiment 1, adults, 11-year-olds, and 8-year-olds rated how well individual test tones continued fragments of melodies. In Experiment 2, 11-, 8-, and 5-year-olds sang continuations to 2-tone stimuli. Response patterns were analyzed using 2 models of melodic expectancy. Despite having fewer predictor variables, the 2-factor model (E. G. Schellenberg, 1997) equaled or surpassed the implication-realization model (E. Narmour, 1990) in predictive accuracy. Listeners of all ages expected the next tone in a melody to be proximate in pitch to the tone heard most recently. Older listeners also expected reversals of pitch direction, specifically for tones that changed direction after a disruption of proximity and for tones that formed symmetric patterns.