www.elsevier.com/locate/cognit This paper argues that a theory of situated vision, suited for the dual purposes of object recognition and the control of action, will have to provide something more than a system that constructs a conceptual representation from visual stimuli: it will also need to provide a special kind of direct (preconceptual, unmediated) connection between elements of a visual representation and certain elements in the world. Like natural language demonstratives (such as `this ' or `that') this direct connection allows entities to be referred to without being categorized or conceptualized. Several reasons are given for why we need such a preconceptual mechanism which individuates and keeps track of several individual objects in the world. One is that early vision must pick out and compute the relation among several individual objects while ignoring their properties. Another is that incrementally computing and updating representations of a dynamic scene requires keeping track of token individuals despite changes in their properties or locations. It is then noted that a mechanism meeting these requirements has already been proposed in order to account for a number of disparate empirical phenomena, including subitizing, search-subset selection and multiple object tracking

Four experiments investigated infants' sensitivity to large, approximate numerosities in auditory sequences. Prior studies provided evidence that 6-month-old infants discriminate large numerosities that differ by a ratio of 2.0, but not 1.5, when presented with arrays of visual forms in which many continuous variables are controlled. The present studies used a head-turn preference procedure to test for infants' numerosity discrimination with auditory sequences designed to control for element duration, sequence duration, interelement interval, and amount of acoustic energy. Six-month-old infants discriminated 16 from 8 sounds but failed to discriminate 12 from 8 sounds, providing evidence that the same 2.0 ratio limits numerosity discrimination in auditory-temporal sequences and visual-spatial arrays. Nine-month-old infants, in contrast, successfully discriminated 12 from 8 sounds, but not 10 from 8 sounds, providing evidence that numerosity discrimination increases in precision over development, prior to the emergence of language or symbolic counting.

In two experiments, a manual search task explored 12- to 14-month-old infants' representations of small sets of objects. In this paradigm, patterns of searching revealed the number of objects infants represented as hidden in an opaque box. In Experiment 1, we obtained the set-size signature of object-file representations: infants succeeded at representing precisely 1, precisely 2, and precisely 3 objects in the box, but failed at representing 4 (or even that 4 is greater than 2). In Experiment 2, we showed that infants' expectations about the contents of the box were based on number of individual objects, and not on a continuous property such as total object volume. These findings support the hypothesis that infants maintained representations of individuals, that object-files were the underlying means of representing these individuals, and that object-file models can be compared via one-to-one correspondence to establish numerical equivalence.

Young infants construct models of the world composed of objects tracked through time and occlusion. To date little is known about the degree to which these models are sensitive to the material makeup of the represented individuals. Two experiments probed 8-month-olds' ability to represent different kinds of entities: rigid, cohesive objects, flexible, cohesive objects, and non-rigid, non-cohesive portions of sand. In Experiment 1, infants represented an array of two rigid, cohesive objects hidden behind a single screen, but failed to represent hidden arrays of two flexible objects or two portions of sand. In Experiment 2, entities were hidden behind two screens instead of one, thereby reducing the information processing demands of the task. In that case, infants succeeded in representing arrays of both types of object stimuli, but again failed to represent the portions of sand. It is argued that (1) the processes by which infants individuate and track entities are sensitive to material kind, (2) rigid cohesive objects occupy a privileged status in this system, and (3) early knowledge about objects and substances has a quantificational aspect. q 2002 Published by Elsevier Science B.V.

Three experiments explored the semantics of the mass-count distinction in young children and adults. In Experiments 1 and 2, the quantity judgments of participants provided evidence that some mass nouns refer to individuals, as such. Participants judged one large portion of stuff to be "more" than three tiny portions for substance-mass nouns (e.g. mustard, ketchup), but chose according to number for count nouns (e.g. shoes, candles) and object-mass nouns (e.g. furniture, jewelry). These results suggest that some mass nouns quantify over individuals, and that therefore reference to individuals does not distinguish count nouns from mass nouns. Thus, Experiments 1 and 2 failed to support the hypothesis that there exist one-to-one mappings between mass-count syntax and semantics for either adults or young children. In Experiment 3, it was found that for mass-count flexible terms (e.g. string, stone) participants based quantity judgments on number when the terms were used with count syntax, but on total amount of stuff when used with mass syntax. Apparently, the presence of discrete physical objects in a scene (e.g. stones) is not sufficient to permit quantity judgments based on number. It is proposed that object-mass nouns (e.g. furniture) can be used to refer to individuals due to lexically specified grammatical features that normally occur in count syntax. Also, we suggest that children learning language parse words that refer to individuals as count nouns unless given morpho-syntactic and referential evidence to the contrary, in which case object-mass nouns are acquired.

We present data from a preferential looking method to investigate when infants have mapped singular and plural markers in English onto the semantic distinction between singleton sets and sets with more than 1 individual. Twenty- to 36-month-old children heard sentences that marked number in 1 of 2 ways: (a) redundantly with verb morphology, lexical quantifiers, and noun morphology (“Look, there ARE SOME blicketS”/“Look, there IS A blicket”) or (b) only with noun morphology (“Look at the blicketS”/“Look at the blicket”). Twenty-four-month-old infants, but not 20-month-old infants, looked at the screen that matched the carrier sentence with respect to singular–plural distinction when number was expressed on the verb, on the noun, and with quantifiers. Detailed looking-time analyses suggest that the arrays begin to be differentiated on the child’s hearing are or is. Twenty-four-month-olds failed when number was marked on the noun alone, whereas 36-month-olds suc-Correspondence should be addressed to Sid Kouider, Laboratoire de Sciences Cognitives et

The violation-of-expectation paradigm investigates infants' physical knowledge by exploiting their tendency to look longer at events that are surprising, unexpected, or physically impossible. The current simulation study examines the role of prediction as a fundamental component of infants' expectations in physical-knowledge studies. A recurrent network is presented with a computer-animated version of Baillargeon’s “car study ” (1986; Baillargeon & DeVos, 1991), in which a car rolls down a ramp and behind a screen. After learning to predict the outcome of a training event, the model is then tested on possible and impossible events from the same study. During testing, the model successfully predicts only superficial features of the test events. These results are used to argue for the necessity of prior physical knowledge, and perhaps also a built-in capacity for mental representation, in order for a prediction system to work.

Much of our interaction with the visual world requires us to isolate some currently important objects from other less important objects. This task becomes more difficult when objects move, or when our field of view moves relative to the world, requiring us to track these objects over space and time. Previous experiments have shown that observers can track a maximum of about 4 moving objects. A natural explanation for this capacity limit is that the visual system is architecturally limited to handling a fixed number of objects at once, a so-called magical number 4 on visual attention. In contrast to this view, Experiment 1 shows that tracking capacity is not fixed. At slow speeds it is possible to track up to 8 objects, and yet there are fast speeds at which only a single object can be tracked. Experiment 2 suggests that that the limit on tracking is related to the spatial resolution of attention. These findings suggest that the number of objects that can be tracked is primarily set by a flexibly allocated resource, which has important implications for the mechanisms of object tracking and for the relationship between object tracking and other cognitive processes.

This study investigates children's ability to use language to guide their choice of individuation criterion in the domains of objects and events. Previous work (Shipley, E. F., & Shepperson, B. (1990). Countable entities: developmental changes. Cognition, 34, 109 -- 136.) has shown that children have a strong bias to use a spatio-temporal individuation strategy when counting objects and that children will ignore a conflicting linguistic description in favor of this spatio-temporal bias. Experiment 1 asked children (3-, 4-, and 5-year-olds) and adults to count objects and events under different linguistic descriptions. In the object task, subjects counted pictures of familiar objects split into multiple pieces (as in Shipley, E. F., & Shepperson, B. (1990). Countable entities: developmental changes. Cognition, 34, 109 -- 136.) and described either using an appropriate kind label (e.g. "car") or the general term "thing". In the event task, subjects watched short animated movies consisting of a goal-oriented event achieved via multiple, temporally separated steps. The events were described either with an appropriate telic predicate targeting the goal (e.g. "paint a flower") or with an atelic predicate targeting the steps in the process (e.g. "paint") and the subjects' task was to count the events. Relative to adults, children preferred a spatio-temporal counting strategy in both tasks; there was no difference among the three groups of children. However, children were able to significantly change their counting strategy to follow the linguistic description in the event but not the object task. Experiment 2 extended the object task to include counting of other types of non-spatio-temporal units such as sub-parts of objects and collections. Results showed that children could ...

Four experiments investigated whether 12-month-old infants use perceptual property information in a complex object individuation task, using the violation-of-expectancy looking time method (Xu, 2002; Xu & Carey, 1996). Infants were shown two objects with di#erent properties emerge and return behind an occluder, one at a time. The occluder was then removed, revealing either two objects (expected outcome, if property di#erences support individuation) or one object (unexpected outcome). In Experiments 1--3, infants failed to use color, size, or a combination of color, size, and pattern di#erences to establish a representation of two distinct objects behind an occluder. In Experiment 4, infants succeeded in using cross-basic-level-kind shape di#erences to establish a representation of two objects but failed to do so using withinbasic -level-kind shape di#erences. Control conditions found that the methods were sensitive. Infants succeeded when provided unambiguous spatiotemporal information for two objects, and they encoded the property di#erences during these experiments. These findings suggest that by 12 months, di#erent properties play di#erent roles in a complex object individuation task. Certain salient shape di#erences enter into the computation of numerical distinctness of objects before other property di#erences such as color or size. Since shape di#erences are often correlated with object kind di#erences, these results converge with others in the literature that suggest that by the end of the first year of life, infants# representational systems begin to distinguish kinds and properties.