The Mental Representation of Sound: A Neural Darwinist Perspective

PLAN:

• some thoughts that grew out of work on auditory/musical expectation
• will present: (1) results from several contrasting experiments; (2) number of disparate facts about hearing; (3) and try to tell a story that makes sense of all these observations.

REPRESENTATION:

• Recognition:
• Imagination:
  
  E.g. sound of an airplane passing overhead
• Anticipation/Expectation:
  
  E.g. Imagine sound of airplane passing overhead.
• Mental Query:
  
  Does the word "but" occur in the lyrics to "Row, Row, Row Your Boat?"
• What do we anticipate in a signal?
  
  
• Brains do not store sounds per se; they interpret, distill, and recast sounds
• When recognizing, imagining, expecting and processing these images, what is represented?

THE USUAL SUSPECTS:

• frequency
• spectral centroid
• tontopy
• firing rate
• absolute pitch
• relative pitch height (high, medium, low)
• pitch chroma
• interval (+M2, -m3 ...)
• scale degree (do, re, mi ...)
• scale degree diad (do-re, re-mi, mi-fa ...)
• contour
• ...

LEARNING:

• Baldwin effect: Pacific bullfrog; salimander
  Learning is a biological phenomenon.
• Plasticity: Patricia Flower's riding accident
• Intersubject variation: e.g. brain imaging (Janata, et al.): people are different (Janata et al.)

NEURAL DARWINISM:

• The wiring of the cortex arises in a manner similar to evolution by natural selection.
• Patterns of neural organization arise spontaneously in response to sensory, somatic, and other inputs.
• Patterns that prove useful are retained and strengthened.
• Patterns that are not useful atrophy.
• Patterns of neural organization compete for cortical real-estate.

WHAT IS "USEFUL"?

• Accurate expectation
• "Primary Affect"

SOME OBSERVATIONS (I): LOCALIZATION:

• Paul Hofman and John Van Opstal (1998). Effect of plastic pinnas on perception of elevation: learned
• Hofman, Vlaming, Termeer, and Van Opstal (2002). Effect of contralateral in-the-ear hearing aids on perception of azimuth: not learned
• Notice that Hofman's experiments are consistent with the Baldwin Effect.

SOME OBSERVATIONS (II): ABSOLUTE PITCH:

• AP: Some have it, and some don't. Why?
• Otto Abraham (1901): "unlearning" theory
• Importance of early exposure
• Evidence of learning: Miyazaki (1989), Simpson & Huron (1994) - consistent with Hick-Hyman law)
• AP as a liability (Miyazaki).

SEVEN FACTS ABOUT AP:

1. Not everyone develops absolute pitch.
2. If absolute pitch emerges, the basis for it is laid in early life.
3. Reaction time data shows that AP is acquired by exposure to the environment -- faster reaction times happen for those pitches that are encountered most often.
4. Possession of AP doesn't mean that the person can only code pitches this way (e.g. scale degree).
5. Nevertheless, possession of absolute pitch can retard the development of relative or intervalic pitch coding (Miyazaki).
6. Absolute pitch proves useless in environments where there is no standard tuning.
7. Absolute pitch never develops in sound environments where it is not useful.

CORRELATED REPRESENTATIONS:

THIRD VARIABLE PROBLEM:

• Average correlation between 3 pitch-related representations: +0.80 !! (interval, interval diad, and scale-degree diad)
• Lots of experiments that purport to show that listeners are sensitive to the manipulation of "X" (where X might be pitch, chroma, interval, contour, etc.). But since these representations are so correlated, one cannot draw such conclusions without careful experimental controls.

INDIVIDUAL DIFFERENCES: CODING MELODIC INTERVALS:

• Do I haear intervals?
• Huron & Aarden experiment:

• Result: Only about 1/3 of participants hear intervals the way I do.
• N.B. W're not all hearing the same thing! Absolute Pitch is not the only way people differ in coding sounds.

CONCLUSION:

1. Some sound representations are physiologically "given" (e.g. tonotopy, firing patterns).
2. Aspects of the acoustical environment that are highly stable (e.g. relationship of azimuth to geometry), appear to lead to innate representations (interaural differences), whereas acoustically variable features (e.g. pinna shape) appear to be learned.
3. At the cortical level, there is considerable brain plasticity.
4. Listeners appear to code sounds in multiple concurrent ways (pitch height, chroma, contour, scale degree, etc.)
5. But listeners differ in which codes they rely on most (e.g. AP).
6. At least in the cases of AP and scale degree, these can only be acquired through exposure to some environment.
7. At least in the case of AP, acquisition depends on (i) whehter it is useful in the environment, (ii) early exposure to that environment.
8. Representations are often highly correlated, so individual differences may not be readily apparent.
9. Although not discussed here, a wealth of evidence from the past decade supports the preeiment role of statistical learning in audition.

That there are competing mental representations for sound.
1. That representations are shaped by exposure to the environment.
2. That representations are differentially favored depending on their predictive success, and on the unique developmental history of the individual.

Obviously, this is all speculative hand-waving: more work is needed.