Auditory object and speech recognition

Question 1

Object recognition

Answer 1

• What is figure and what is background?
  
  • Contrast signals that define an object as opposed to visual background
  • Recognizing an object (e.g. word, animal vocalization, action (footsteps, breaking glass))
• Viewpoint invariance
  
  • Frequency and time domain patterns independent of actual frequency content
  • Recognition independent of spatial RF of particular cells in early visual system
• Semantic aspect, perceptual recognition: what is it?

Question 2

Visual and auditory objects

Answer 2

• There is an inherent temporal component to the definition of an auditory object
  
  • Auditory objects are defined in the time and frequency domain
• Visual objects are defined by spatial luminance and chromatic contrast
  
  • Much information about a visual scene can be taken in ‘at a glance’
    
    • E.g., Thorpe et al (1996): identification of animal (or not) present in visual scene with 10 ms presentation

Question 3

Cues for segmentation

Answer 3

• Identification involves aligning sensory information with sources
• Spectral cues
  
  • Pitch
  • Timbre
• Temporal cues
  
  • Onset time (<30 ms)

Question 4

Dorsal and ventral processing streams in auditory cortex suggested from animal experiments

Answer 4

• Evidence for more spatial responses in dorsal areas seems quite strong, but selectivity of ventral areas for ‘what’ over where has been questioned
• Proposal based on language processing: ventral stream for perception, dorsal for production

Question 5

Categorical perception of consonant- vowel syllables

Answer 5

• Synthetic CV stimuli varying continuously from /ba to /da to /ga
• Perception changes abruptly (categorically)

Question 6

Discuss the neural correlates of categorical perception

Answer 6

see pics