Cognitive Neuroscience Lecture 10: Auditory Perception

''L10: Auditory Perception ''

The challenges of audition

-         Complex auditory scene analysis

o How do we recognize identity and location of objs in complex scenes?

Overview

1.      Ear to brain

2.      Neural representation of intensity, frequency, and location

3.      Stream segregation

a. Grouping principles

4.      Top-down processing

a. Inference

Auditory object recognition

-         What (obj recognition) requires:

o Invariance (obj constancy): what each obj shares with other members of category (diff dogs vs diff cows)

o Specificity: what is specific to each object (e.g. one dog bark vs another)

o Have to solve both of these at the same time

-         Where (sound location is not given by geometry of sensory surface like it is for vision)

o Sound: audible variations in air pressure (mvt of air molecs)

o Frequency: number of compressed air molecs that pass by ears/sec; pitch

o Cycle: wavelength

o Hertz: s-1 (humans respond to 20-20,000 Hz)

o Amplitude: height of peak; volume

o Complex sounds can be decomposed into constituent freqs and amps

-         Challenge: extracting freqs and amps

o Tympanic membraneà ossiclesàoval windowà fluid in cochleaàbasilar membraneàsensory neurons

-         Basilar membrane coiled inside of cochlea

o Base narrow and rigid: high freq

o Apex wide and floppy: low freq

o As frequency decreases, maximal deformation moves toward apex

-         Tonotopic organization of basilar membrane

o Tonotopic map: adj tones in adj locations of neural surface

o Bending basilar membraneàrelease NT

o Activates spiral ganglion cells àaction potentials

§ Spiral ganglion cells: axons form auditory nerve (VIII)

-         Auditory pathways

o Multiple paths

§ Ascending converge on inferior colliculus

§ Splits

o Bilateral connectivity

§ Superior olivary has info from both ears

§ All nuclei above cochlear nuclei receive input from both ears (although stronger contralateral)

§ (cochlear nuclei are in SC)

o Extensive feedback

§ A1 projects back to MGN and inf. Coll.

§ Considerable processing prior to cortex

-         Full pathway:

o   Auditory nerveàcochlear nucleusàsuperior oliveàinferior colliculusàMGNàA1

o   Tympanic membraneà ossiclesàoval windowà fluid in cochleaàbasilar membraneàsensory neuronsàauditory nerve fiber(spiral ganglion cells)àipsilateral cochlear nucleusàsuperior olivary nucleusàinferior colliculus à medial geniculate nucleusà contralateral auditory cortex

-         '''Auditory cortex: superior temporal gyrus '''

o Heschl’s gyrus: primary auditory cortex (A1) (core)

o Planum temporale: 2ndary auditory cortex (belt)

o Core, belt, parabeltà increasingly complex/associative representation and processing

Neural representation of features of sound

-         Intensity (amplitude) (loudness)

o Firing rate

§ Greater amplitude of vibration of basilar membraneàgreater number of AP in spiral ganglion cells

o Number of active neurons

§ Greater amplitude of vibrations of basilar membraneàlarger area of basilar membrane affectedàgreater number of SGC firing

-         Frequency

o Local distortion of basilar membà activation of specific spiral ganglion neurons

o '''Frequency tuned neurons''': each respond maximally to specific freqs; fairly sharp, not as sharp as visual

Auditory cortex: Tonotopy

-         Sylvian fissure

o Apex of cochlea/lower frequency corresponds to anterior A1

o Base/high freq is posterior a1

Horizontal location ‘where’

-         Interaural time difference

o Interaural time delay = difference between one ear and another = 0.6 ms

o Encoding difference = determining horizontal location

-         Interaural intensity difference

o Head casts a sound shadow creating differences in intensity of the sound in each ear

o ID location of a sound source

-         Computing ITD

o Brainstem neurons ‘prefer’ (max firing) for certain ITD

o Key step: action potentials converge on an MSO neuron that responds most strongly if their arrival is a certain ITD

o Longer path from sound side to non-sound side than from sound side to its own side

-         Computing IID

o LSO

o Different locations of sound sourceà different patterns of excitation and inhibition; characteristic signals depending on location of sound

o Stronger stim to one side excites that side LSO, and inhibits opposite LSO

o Excitation from sound’s side greater than inhibition from right side à net excitation on sound side to higher centers

-         '''ITD=MSO, IID=LSO'''

-         Auditory perception

o Audition is not a simple mirroring of stim; influenced by knowledge, and it influences other things

o Reveals inner workings of prediction engine

-         Stream segregation

o Auditory Inference: grouping by similarity and proximity (Gestalt principles)

§ Group by similarity in frequency if it’s fast enough

§ One melody for high freq, one melody for low freq

§ “assumes” that similar sounds are more likely to come from same object (esp if close in time)

o Grouping by common fate

§ During modulation, hear separate sounds

§ After modulation, tones come back together

§ Auditory system assumes freq that fluctuate together are more likely to come from same objs than those that do not

o Object knowledge influences perception: top-down processing

§ Perceptual filling in; “explanation for the gap” if noise louder than original sound

§ Object knowledge exerts top-down effectàinfluences lower-level processes (filling in)

§ Similar to vision amodal completion

1.      ID freq, locations, intensities

2.      Group into possible objs (e.g. similarity)

3.      Auditory ID (requires obj knowledge (learned); sound of something you already know)

Bottom up = 1, 2, 3

Top-down = 3, 2, 1