Elsevier

Cognition

Volume 122, Issue 1, January 2012, Pages 51-60
Cognition

Stimulus-dependent flexibility in non-human auditory pitch processing

https://doi.org/10.1016/j.cognition.2011.08.008Get rights and content

Abstract

Songbirds and humans share many parallels in vocal learning and auditory sequence processing. However, the two groups differ notably in their abilities to recognize acoustic sequences shifted in absolute pitch (pitch height). Whereas humans maintain accurate recognition of words or melodies over large pitch height changes, songbirds are comparatively much poorer at recognizing pitch-shifted tone sequences. This apparent disparity may reflect fundamental differences in the neural mechanisms underlying the representation of sound in songbirds. Alternatively, because non-human studies have used sine-tone stimuli almost exclusively, tolerance to pitch height changes in the context of natural signals may be underestimated. Here, we show that European starlings, a species of songbird, can maintain accurate recognition of the songs of other starlings when the pitch of those songs is shifted by as much as ±40%. We observed accurate recognition even for songs pitch-shifted well outside the range of frequencies used during training, and even though much smaller pitch shifts in conspecific songs are easily detected. With similar training using human piano melodies, recognition of the pitch-shifted melodies is very limited. These results demonstrate that non-human pitch processing is more flexible than previously thought and that the flexibility in pitch processing strategy is stimulus dependent.

Introduction

Periodicity is a fundamental property of many natural sounds, and its perception, “pitch”, plays a central role in auditory processing in both humans and animals. Although humans are capable of using absolute pitch height in some auditory tasks, the intervals between sounds, referred to as “relative pitch”, are generally more important in sound recognition (Attneave & Olson, 1971). In contrast, most animals are thought to rely primarily on absolute pitch for auditory recognition (Hulse and Cynx, 1985, Hulse et al., 1984, Page et al., 1989). These comparative studies have focused almost exclusively, however, on stimuli constructed from sine-wave tones. Here, using songbirds as a model, we contradict this prevailing view by showing in a series of studies that non-human pitch processing strategies are flexible and stimulus-dependent. Over similar ranges of shifts in absolute pitch, recognition of conspecific songs is maintained, but recognition of tonal melodies is quite poor.

Among normal adult humans, the ability to reliably recognize pitches without an external reference is rare. Even musically trained individuals with excellent relative pitch often cannot provide an accurate note label (i.e. C, F#) for individual pitches (Ward, 1999). Among the general population approximately 1 in 10,000 (Ward, 1999) have this ability, known as musical absolute pitch (musical AP). In this paper, we use the term absolute pitch not to refer to this music-specific form of note recognition, but to memory for pitch height, which corresponds to the fundamental frequency of a sound.

Long-term memory for pitch height appears to be common, even among individuals without musical AP (Schellenberg & Trehub, 2003). For example, when asked to hum or sing a very familiar song, the median of the distribution of people’s starting pitches is broad, but centered close to the original source (Levitin, 1994). Likewise, many individuals without musical AP can correctly identify when versions of familiar television theme songs are transposed from their original pitch (Schellenberg & Trehub, 2003). Furthermore, musically untrained listeners (without AP) can identify whether a dial tone is at the correct pitch, too high or too low, even though they do not possess musical AP and are unable to attach specific pitch labels to the notes they recognize or produce (Smith & Schmuckler, 2008).

A widespread feature of human pitch perception is that humans easily recognize two sequences of notes as the same melody if the relationship between pitches is preserved, even if all of the pitches are different (Attneave and Olson, 1971, Dowling and Harwood, 1986). This ability is known as relative pitch, and is observed even in infants (Plantinga & Trainor, 2005). But the ability to recognize sounds independent of their absolute frequency content is more general in human cognition. Normal listeners have no difficulty understanding the same word spoken by two different individuals, even in a tonal language, because it is the pitch contour and relationship between pitches rather than the absolute frequency of words that are used to convey lexical and intonational meaning (Ladd, 2008).

Examples of relative pitch perception among non-human animals are rare. In fact, outside of human music perception, the only documented cases of relative pitch perception involve recognizing a single interval (Hurley et al., 1992, Shackleton et al., 1992, Weisman et al., 1990, Yin et al., 2010). Relative pitch plays a role in the calls and song of some avian species, but again only a single interval between two notes has been implicated (Christie et al., 2004, Weisman et al., 1990). For example, chickadees shift the absolute pitch of their song but preserve the relative pitch between two notes (Shackleton et al., 1992). Recognizing a sequence of three or more pitches by the configuration of their intervals has not been reported in any of these cases. Rhesus monkeys do appear to recognize transposed tonal melodies, but only when transposed by octaves, not by 0.5 or 1.5 octaves (Wright, Rivera, Hulse, Shyan, & Neiworth, 2000).

Previous laboratory studies support a diminished role for relative pitch processing in songbirds and suggest instead that absolute pitch height processing is the dominant strategy (Cynx, 1995, Hulse and Cynx, 1985, Hulse and MacDougall-Shackleton, 1996, Nagel et al., 2010, Page et al., 1989, Weisman et al., 1998, Weisman et al., 2010). In a striking example of their limitation using relative pitch cues, European starlings never successfully learned to recognize ascending and descending tone sequences within a large stimulus set even after 10,000–30,000 training trials (Page et al., 1989). Failure to learn this recognition suggests an inability to use relative pitch when absolute pitch cues were not informative. Recognizing whether a tone sequences is ascending or descending would be trivial for most human subjects.

If a smaller stimulus set is used, starlings eventually learn to recognize ascending and descending tone sequences after many trials. However, when presented with stimuli outside of the pitch range used during training, subjects took as long to learn to accurately recognize the new stimuli as they did to learn the original stimuli (Cynx, 1995, Hulse and Cynx, 1985). Indeed, in all cases where pitch generalization has been observed, it is constrained by the absolute pitch range of the training examples (Cynx et al., 1986, Hulse and Cynx, 1985). Hulse termed this property the “frequency range constraint”.

In contrast to their deficit recognizing stimuli on the basis of relative pitch, many species accurately use absolute pitch height cues. Several bird species, including zebra finches and chickadees perform well in pitch range discrimination studies (e.g. Lee et al., 2006, Weisman et al., 1998). In fact, these studies have repeatedly observed better encoding of pitch height than is observed in most humans (Weisman et al., 2010). Zebra finches (Taeniopygia guttata) are also able to detect mistunings in harmonic complexes with lower thresholds than humans, suggesting that they have very fine-grained pitch perception abilities (Lohr & Dooling, 1998). Thus, although songbirds can use both relative and absolute pitch cues for recognition, their ability to use relative pitch appears to be very limited, secondary to absolute pitch cues, and to require extensive training.

Does the observed bias toward absolute pitch cues in tone sequence recognition reflect general perceptual limitations in the ability to recognize stimuli shifted in pitch? To investigate this question, we studied pitch flexibility in a series of recognition tasks by European starlings, where a detailed set of prior experiments has established a bias toward absolute pitch cues in tone sequence recognition. We examined whether starlings continue to recognize their own species’ song in the context of changing pitch. Are they committed to absolute pitch as a recognition cue, or are they more flexible?

We conducted operant training experiments using stimuli derived from conspecific song. In the first experiment, we trained two groups (of four subjects each) to recognize conspecific song excerpts. The first group learned to recognize songs that were always played at the same pitch, so that absolute frequency cues were useful for song recognition. The second group was trained with the same set of songs shifted to cover a range of absolute pitch levels. After training, we tested their ability to maintain recognition performance with these songs at novel pitch levels. In a second experiment, we tested whether two subjects could explicitly use pitch height cues in the context of song recognition by training subjects to recognize high pitch and low pitch versions of the same song, and then measuring recognition of songs at intermediate pitches. As a control, we also trained three starlings to recognize two tonal melodies, and then tested the generalization of this learned recognition to pitch shifted versions of the same melodies.

Section snippets

Subjects

Eleven European starlings (Sturnus vulgaris) of unknown sex served as subjects in these experiments. Previous operant song recognition studies (e.g. Gentner, Hulse, Ball, & Bentley, 2000) have not revealed any measurable behavioral differences in performance between male and female subjects. All subjects were captured as adults in Southern California between 2008 and 2010. After capture, subjects were housed in a mixed-sex aviary. The photoperiod in both the aviary and experimental apparatus

Starlings rapidly learn to recognize pitch-shifted song

In experiment 1, we trained eight subjects to recognize excerpts of starling song. Four subjects learned four unshifted training exemplars, and the other four subjects learned 64 shifted exemplars. Recognition accuracy improved rapidly with training, with subjects reaching a d-prime of 1 in an average of 6 100-trial blocks when trained on the unshifted stimulus set, and 10 blocks when trained on the shifted stimulus set (Fig. 2). After 10 blocks of training, recognition performance reached a

General discussion

While humans maintain accurate recognition of words or melodies over large pitch height changes, songbirds and other non-human animals have often been characterized as having difficulty performing such tasks. Perhaps this reflects fundamental differences in the structure or flexibility of songbird auditory representations. Alternatively, because non-human studies have used sine-wave stimuli almost exclusively, tolerance to pitch height changes in the context of more natural signals may be

Acknowledgments

Research supported by NIH/NIDCD R01DC008358 to Timothy Q. Gentner. Aniruddh D. Patel was supported by Neurosciences Research Foundation as part of its program on music and the brain at The Neurosciences Institute, where he is the Esther J. Burnham Senior Fellow. The authors thank three anonymous reviewers for helpful suggestions.

References (43)

  • J. Plantinga et al.

    Memory for melody: Infants use a relative pitch code

    Cognition

    (2005)
  • O. Tchernichovski et al.

    A procedure for an automated measurement of song similarity

    Animal Behaviour

    (2000)
  • R.G. Weisman et al.

    Absolute pitch in boreal chickadees and humans: Exceptions that test a phylogenetic rule

    Learning and Motivation

    (2010)
  • F. Attneave et al.

    Pitch as a medium: A new approach to psychophysical scaling

    American Journal of Psychology

    (1971)
  • B. Chandrasekaran et al.

    Individual variability in cue-weighting and lexical tone learning

    Journal of the Acoustical Society of America

    (2010)
  • P.J. Christie et al.

    Pitch shifts and song structure indicate male quality in the dawn chorus of black-capped chickadees

    Behavioral Ecology and Sociobiology

    (2004)
  • Creel, S. C., & Tumlin, M. A. (in press). On-line recognition of music is influenced by relative and absolute pitch...
  • J. Cynx

    Similarities in absolute and relative pitch perception in songbirds (starling and zebra finch) and a nonsongbird (pigeon)

    Journal of Comparative Psychology

    (1995)
  • J. Cynx et al.

    A psychophysical measure of pitch discrimination loss resulting from a frequency range constraint in European starlings (Sturnus vulgaris)

    Journal of Experimental Psychology: Animal Behavior Processes

    (1986)
  • M. Dolson

    The phase vocoder: A tutorial

    Computer Music Journal

    (1986)
  • W.J. Dowling et al.

    Music cognition

    (1986)
  • T.Q. Gentner

    Temporal scales of auditory objects underlying birdsong vocal recognition

    Journal of the Acoustical Society of America

    (2008)
  • T.Q. Gentner et al.

    Individual vocal recognition and the effect of partial lesions to HVc on discrimination, learning, and categorization of conspecific song in adult songbirds

    Journal of Neurobiology

    (2000)
  • D.M. Green et al.

    Signal detection theory and psychophysics

    (1966)
  • S.L. Greenspan et al.

    Perceptual learning of synthetic speech produced by rule

    Journal of Experimental Psychology: Learning, Memory, and Cognition

    (1988)
  • L.L. Holt et al.

    Cue weighting in auditory categorization: Implications for first and second language acquisition

    Journal of the Acoustical Society of America

    (2006)
  • S.H. Hulse et al.

    Relative pitch perception is constrained by absolute pitch in songbirds (Mimus, Molothrus, and Sturnus)

    Journal of Comparative Psychology

    (1985)
  • S.H. Hulse et al.

    Absolute and relative pitch discrimination in serial pitch perception by birds

    Journal of Experimental Psychology: General

    (1984)
  • S.H. Hulse et al.

    Concurrent absolute and relative pitch processing by European starlings (Sturnus vulgaris)

    Journal of Comparative Psychology

    (1996)
  • T. Hurley et al.

    Relative pitch recognition in white-throated sparrows

    Animal Behaviour

    (1992)
  • P. Iverson et al.

    A perceptual interference account of acquisition difficulties for non-native phonemes

    Cognition

    (2003)
  • Cited by (26)

    • Music for animal welfare: A critical review & conceptual framework

      2022, Applied Animal Behaviour Science
      Citation Excerpt :

      Compared to humans and rats, birds are superior at discriminating pure tones from different frequency ranges (Weisman et al., 2012). But while humans can easily recognize a melody that has been transposed a few semi-tones as the same melody (which relies on relative pitch perception), zebra finches did not recognise songs when the frequencies were shifted up or down by 8% or more ((Nagel et al., 2010 but see Bregman et al., 2012 for better performance in starlings). Animals also differ from humans in perception of timbre (with black-capped chickadees showing poor generalization to novel timbres; Hoeschele et al., 2012), octave equivalence (being absent in chickadees and budgerigars; Hoeschele et al., 2013; Wagner et al., 2019), and consonance (showing mixed results; Hulse et al., 1995; Izumi, 2000; McDermott and Hauser, 2004; Wagner et al., 2020).

    • Prosody in birdsong: A review and perspective

      2017, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      This seems to be in contrast with pitch perception in human speech (i.e. perception of lexical tone and intonation), which is generally based on relative pitch rather than absolute pitch perception, since it can be generalized across speakers (see also Section 3.2). However, the birds’ generalization of pitch-shifted stimuli seems to be influenced by the type of stimulus, in that generalization might be better across more natural sounds than across artificial sounds (Bregman et al., 2012). Furthermore, a recent study suggests that starlings might rely on the shape of the spectral envelope, and not on pitch, for tone sequence recognition (Bregman et al., 2016).

    • Is Birdsong More Like Speech or Music?

      2016, Trends in Cognitive Sciences
    • Gradient language dominance affects talker learning

      2014, Cognition
      Citation Excerpt :

      Participants were individually tested in a sound isolated room, and audio was presented using Sennheiser HD 280 Pro headphones, which participants could adjust to a comfortable loudness level. We designed this experiment to follow a set of training and testing procedures that have been widely applied in animal learning studies (e.g. Bregman et al., 2012; Gentner and Hulse, 1998; Hulse and Dorsky, 1979; Scharff et al., 1998) and human learning studies (e.g. Anderson, 1976; Gathercole and Baddeley, 1990; Reber, 1967). In these studies, participants are trained for a variable amount of time until they reach a criterion level of performance, and are then tested on novel exemplars to verify accurate generalization.

    • Processing of communication sounds: Contributions of learning, memory, and experience

      2013, Hearing Research
      Citation Excerpt :

      Moreover, in a set of studies by Petersen and colleagues (Beecher et al., 1979; Petersen et al., 1978, 1984; Zoloth et al., 1979), Japanese macaques more easily learned to discriminate sets of CVs that were divided by a communicatively-relevant dimension (the position of a frequency-inflection peak), rather than by an arbitrary one (initial pitch). Similarly, European starlings are capable of learning to recognize pitch-shifted conspecific songs, but not pitch-shifted piano melodies (Bregman et al., 2012). A mnemonic advantage for CVs has also been reported in two recent studies of auditory short-term memory in humans (Weiss et al., 2012) and non-human primates (Ng et al., 2009).

    • Perceptual categories enable pattern generalization in songbirds

      2013, Cognition
      Citation Excerpt :

      It is unclear whether successful pattern generalization was enabled because the perceptual features of motifs aligned with already known natural categories of song elements i.e. warble and rattle, or because the categorical boundaries were more easily acquired during training due to greater within-class perceptual similarities, or both. Our choice to use of species-typical song elements was motivated by recent demonstrations that such stimuli, compared to simple tones, enhance task performance on other auditory tasks (Bregman, Patel, & Gentner, 2012). We suspect that starlings would perform similarly on this same pattern task using other auditory objects so long as they are sensitive to the perceptual categories of those stimuli – this remains, however, an open question for future research.

    View all citing articles on Scopus
    View full text