Professional Documents
Culture Documents
Infants Posses A System of Numerical Knowledge Karen Wynn
Infants Posses A System of Numerical Knowledge Karen Wynn
Infants Posses A System of Numerical Knowledge Karen Wynn
able representations of small num- moved from a collection of two ob- presented with 1 -I- 1 situations, and
bers, and that these representations jects. Infants in the / -H / group saw another group was presented with 2
are independent of the perceptual one item placed into a display case. - 1 situations; for both groups, the
properties of specific arrays. A screen then rotated up to hide the outcome was sometimes one object
item, and the experimenter brought and sometimes two objects. All the
a second item into the display and objects were placed on a large re-
placed it out of sight behind the volving plate in the center of the dis-
INFANTS HAVE screen (Fig. 2, top). The 2 - 1 group play, which was occluded when the
PROCEDURES THAT
saw two items placed into the dis- screen was raised. The objects were
SUPPORT
play. After the screen rotated up to therefore in continuous motion, so
NUMERICAL REASONING
hide them, the experimenter's hand no object retained a distinct spatial
reentered the display, went behind location throughout the experimen-
Possessing genuine numerical the screen, and removed one item tal operation. Nonetheless, infants
knowledge entails more than simply from the display (Fig. 2, bottom). For looked reliably longer at the numer-
the ability to represent different both groups, the screen then ically incorrect outcomes, showing
numbers. A numerical system is dropped to reveal either one or two that they were computing over the
composed not only of numbers, but items. Infants' looking times to the number of objects, not over the
also of procedures for manipulating displays were then recorded. filled-or-empty status of different
these numbers to yield further nu- Pretest trials, in which infants spatial locations.^
merical information. Infants might were simply presented with displays
be able to determine numbers of en- of one and two items to look at, re-
tities without being able to reason vealed no significant preference for
about these numbers or to use them one number over the other, and no A MECHANISM FOR
to make numerical kinds of infer- significant difference in preference DETERMINING AND
ences. If so, we would not want to between the two groups. But there REASONING
credit infants with a system of nu- was a significant difference in the ABOUT NUMBER I
merical knowledge. looking patterns of the two groups
Studies conducted in my labora- on the test trials: Infants in the 1 + 1 The ability to discriminate small
tory show that 5-month-old infants group looked longer at the result of numbers of entities precisely, and in
are able to engage in numerical rea- one item than the result of two some cases to perform numerical
soning: They have procedures for items; infants in the 2 — 1 situation operations over these numbers, has
manipulating their numerical repre- looked longer at the result of two been documented in a variety of
sentations to obtain information of items than the result of one item warm-blooded vertebrate species as
the relationships that hold between (Fig. 3). well as in human infants. This sug-
them. In these experiments, the in- In another experiment, infants gests that a common mechanism
fant is shown a small collection of were shown an addition of one item may have evolved to perform this
objects, which then has an object to another, and the final number of function at a distant point in evolu-
added to or removed from it. The objects revealed was either two or tionary history/
resulting collection of objects that is three. Again, infants looked signifi- The accumulator is a model of a
subsequently shown to the infant is cantly longer at the inconsistent out- mental mechanism for representing
either numerically consistent or in- come of three objects than at the number, originally proposed to ac-
consistent with the addition or sub- consistent outcome of two objects count for numerical abilities in rats.^
traction. Because infants look longer (Fig. 4). (Pretest trials revealed no The accumulator mechanism can
at outcomes that violate their expec- baseline preference to look at three account for both the ability to repre-
tations, if they are anticipating the items over two items.) sent number and the ability to oper-
numberof objects that should result, These results are robust; they ate over these numerical representa-
they will look longer at inconsistent have been obtained in other labora- tions.^ This mechanism produces
outcomes than at consistent ones. tories, using different stimuli and pulses at a constant rate; these
In the first experiment," one with variations in the procedure.^ pulses can be passed into an accu-
group of 5-month-otd infants was One study tested the possibility that mulator by the closing of a switch.
shown an addition situation in infants were anticipating certain spa- For each entity that is to be counted,
which one object was added to an- tial locations to be filled and others the switch closes for a fixed brief in-
other identical object, and another empty, rather than anticipating the terval, passing the pulses into the ac-
group was shown a subtraction situ- number of items in the display. One cumulator during that interval. Thus,
ation in which one object was re- group of 5-month-old infants was the accumulator fills up in equal in-
"1 15 - %^\
\
15 -
Look ing 1 ime
\ 15 • \
ooking 'ime
ooking T me
V
N
10 -
- • 2 dolls dolls
10 - 10 •
T5 1 doll
-° 1 doll
•• 2 dolis ^ " • " ' ^ ^ 2 dolls
5-
5- _j 5-
Test Trials
Test Trials Test Trials
Fig. 4. Five-month-olds' looking times
Fig. 3. Five-month-olds' looking times to outcomes of one doll and two dolls following to two versus three dolls after viewing a
event sequences in which one doll was added to a display of one doll (a) or one doll was sequence of events in which one doll
taken av^-ay from a display of two dolls (b). was added to a display of one doll.
soning in young infants: Seeking explanations for infants, manuscript submitted for publication 320-3J4 (1983): see also C,R. Gallistel and R, Cel-
impossibie events, British journal of Developmental (1995). man, Preverbal and verbal counting and computa-
Psychology. 12, 9-33 (1994); D.S. Moore, Infant tion. Cognition, 44. 43-74 (1992).
7. For review and discussion, see chapter 10 of
mathematical skills? A conceptual replication and 9. See, e.g., R. Gelman, Epigenetic foundations
C.R. Gallistel, The Organization of iearn/ng (MIT
consideration of interpretation, manuscript submit- of knowledge structures: Initial and transcendent
Press, Cambridge, MA, 1990).
ted (or publication (1995); T,|. Simon, SJ. Hespos, constructions, in The Epigenesis of Mind: Essays on
and P. Rochat, Do infants understand simple arilh- 8. For a detailed description of the accumulator Biology and Cognition. S. Carey and R. Gelman,
metic: A replication of Wynn (1992), Cognitive De- model and of experimental support for tl, see W.H. Eds. (Eribaum, Hiilsdaie, Nj, 1991); M, Kline, Math-
velopment, W. 253-269 (1995). Meek and R.M. Church, A mode control model of ematical Thought From Ancient to Modern Times.
6, E. Koechlin, S. Dehaene, and |. Mehler, Nu- counting and timing processes, journal of Experi- Vol. i (Oxford University Press, Oxford, England.
merical transformations in five moath old human mental Psychology: Animal Behayior Processes, 9, 1972),