[top of page 14 of handout]
Morton, J. and Johnson, M.H. (1991) CONSPEC and CONLERN: A two-process theory of infant face recognition. Psychological Review, 98, 164-181.
Evidence from newborns leads to the conclusion that infants are born with some information about the structure of faces. The structural information, termed CONSPEC, guides the preference for facelike patterns found in newborn infants. CONSPEC is constrasted with a device termed CONLERN, which is responsible for learning about the visual characteristics of conspecifics. In the human infant, CONLERN does not influence looking behaviour until 2 months of age. The distinction between these two 2 independent mechanisms allows a reconciliation of the conflicting data on the development of face recognition in human infants. Finally, evidence from another species, the domestic chick, for which a similar 2-process theory has already been put forward, is discussed. The new nomenclature is applied to the chick and used as a basis for comparison with the infant.
Exptl details are in Johnson et al 1991 Cognition paper:
1. Support for the idea of Fanz (1961) that attention is focussed on “stimuli which will later have adaptive significance.” p164
2. For 20 years this has been rejected: infants said to discern only the grossest features, and to prefer the exterior boundaries over internal features.
3 Only at 2 months do they look longer at face-like rather than scrambled pictures. Some data suggest only at 4 months. Habituation techniques are also used.
4. But newborns track “face” stimuli with head and eye movements. e.g. Expt (in Cognition) with 24 infants only 37 mins old (mean). Stimuli either face, scrambled or blank. Both eye and head turn were greatest for the face and least for the blank.
5. The basis of the newborns’ preference may be very general, possibly due to sensory properties of the visual system. The simplest hypothesis is that the innate specification of faces resembles the second panel in Figure 7 (uniform little squares in a circle).
6. 2-month-old infants make differential responses to schematic faces under certain procedures, but 1-month-old infants do not. (p.175).
7. At 5-months, infants show a preference for schematic over scrambled faces if the internal features are moving, but not if they are static. (p. 173-4: by this time, it is assumed some learning about faces has taken place.)
[bottom of page 14 of handout]THE COMBINED THEORY (p.175)
There is some innate specification which directs attention to faces. The effects of this last up to about 1 month, in the peripheral visual field. There is also a learning mechanism, which is non-specialized, but which acquires knowledge about faces, which takes over from the innate mechanism during the first 2 months, and influence behaviour thereafter.
There is general agreement that neonatal perception is sub-cortical, and that cortical mechanism become more dominant after the first 2 months. Thus the innate specification (CONSPEC) is sub-cortical, and the later learning (CONLERN) is cortical.
APPLIED TO CHICKS
It has also been proposed that there are two neural mechanisms used for imprinting in the chick. There is an early predisposition to attend to objects generally similar to conspecifics. Learning about the details of the object exposed to depends on a particular forebrain region (IMHV, p.177. This is a long-term store in the left hemisphere and a buffer store in the right - Horn, 1985.)
IMHV == Intermediate and Medial portions of the Hyperstriatum Ventrale
What happend with the figures shown in the handouts?
[not in handout, see intranet]pp171-172 Figure 7
Simultaneous choice paradigm with 48 infants mean age 1.7 days
There are various theories about amplitude and phase spectra. (Fourier Transformations)
But with all pairs, infants ”overwhelmingly preferred the face” --69% of the time.
“We conclude that the human CONSPEC contains structural information, that is, some specification of the features of a face together with their relative location in space.”
(although they say that they have found greater orientation to a checkerboard than to a face!!)
Goren et al 1975, found that newbords turned further to track faces than scrambled faces, when they were 9 minutes old (they replicated at 40 minutes) In Goren, delivery staff and experimenter wore partial face masks.
[not in handout, see intranet]Figure 8
41 newborns at University College Obstetrics Department between 15 and 69 minutes after birth. Like Expt 1, but infants rested on a holder in the E’s lap. Eye movement but not head turning was significantly greater for the Face.
attempt to replicate Goren, mean age 37 Minutes. Infant on back on E’s lap. E did not know which stimulus being used. Stimulus presented at midline of a “protractor” and moved slowly to one side (5 deg per sec). Video recordings analysed by two independent “blind” observers. Concordance = .87. Both eye and head turning in order for face, scrambled face and blank.
Ex 1. Newborns show preference in face, scrambled face and blank models (both eye and head, Fig 2). But this seems to be lost at 5 weeks
“Strengthens the argument that fairly complex perceptual organization is present at birth (but they say simplest hypothesis is for 2nd panel of figure 7 - just some degree of symmetry and regularity”.
Expt 3 the Rotating Chair at 5 weeks
Stimuli in middle, motor turns chair and infants turn their head and eyes to keep stimulus in view.
Data indicate preference for face but this declines at 19 weeks.
CONLERN is a nonspecialized learnig mechanism. Effectively it learns about the charactersitics of faces because the infant pays a lot of attention to them
More recent work on infants less than 6 months old has tended to confirm eye contact and eye position (upright as opposed to inverted faces) are influential aspects of newborns' preferences for face-like stimuli. e.g. “Eye contact is one of the strongest communicative signals in humans, and it is plausible to assume that evolutionary selection has made human newborns sensitive to such a signal.” (Farroni et al., 2005)