object + face recognition

binding problem

sensory information coming from different channels is brought together + unified to form a coherent stream

feature binding

bind features into objects

space binding

bind objects in space into some arrangement

marr’s theory of object recognition

recognition occurs in 3 stages: the primal sketch, the 2 ½D sketch, the 3D model representation

the primal sketch

initial stage of edge + sketch detection without object recognition

the 2 ½D sketch

visual information includes depth + surface orientation but still dependent on perspective

the 3D model representation

visual system constructs a full 3D model of object

biederman (1987) recognition-by-components theory

we recognise objects by identifying their basic components (geons)

geons

3d shapes that enable us to mentally represent the objects that we easily recognise

pareidolia

phenomenon of perceiving random stimuli as faces

part-whole effect

memory for a face part is more accurate when shown within the face rather than on its own

distortion / inversion effect

faces are harder to recognise when presented inverted due to holistic processing being disrupted

the composite effect

difficulty to detect faces when two separate images are composed together since the brain tries to view them as a whole

the fusiform face area

exhibits a greater response to faces than other objects

the extrastriate body area

responds preferentially to images of human bodies + body parts compared to various inanimate objects

the parahippocampal place area

processes visual information related to places in the local environment

lateral occipital cortex

responsive to shapes

visual agnosia

modality-specific inability to recognise a stimulus that is not attributable to sensory, attentional, linguistic or memory disorders

apperceptive agnosia

can process + judge basic aspects of objects but are unable to combine these into objects / meaningful wholes

associative agnosia

can perceive + integrate visual information to form holistic representations but cannot be linked to stored knowledge

prosopagnosia

visual agnosia for faces, lost ability to recognise a particular face as belonging to an individual person

viewpoint-invariant

recognition works even if the object is seen from a new angle, as long as the geons + their relationships stay visible