Many studies carried out over the last 30 years lead to the suggestion that flexible skills and belief systems that adults employ to process objects, agents, space and numbers might gradually emerge as the results of the interaction between the structure of the input provided by the species-typical environment and initial, innately specified constraints (de Schonen, 2002; Elman et al., 1996; Karmiloff-Smith, 1992, Nelson 2001; Simion, Macchi Cassia, Turati, & Valenza, 2001). Indeed, despite newborns' poor visual acuity, investigators of the earliest beginnings of cognition have come to recognize that from birth, newborns possess very sophisticated attentive (Farroni, Simion, Umiltà & Dalla Barba, 1999; Simion, Valenza & Umiltà, 1995; Valenza, Simion & Umiltà, 1994) and perceptual abilities (Farroni, Valenza, Simion & Umiltà, 2000; Macchi Cassia, Simion, Milani & Umilta`, 2002; Valenza, Leo, Gava & Simion, 2006; Valenza & Bulf, 2007) that allow them to process and represent different kinds of information.
In line with this theoretical framework the present study has addressed to investigate the origin of numerical knowledge
My study begins reporting the language-independent processes and representation systems that allow adults to elaborate numerical information (Chapter 1). Subsequently, the documentation of studies demonstrating that very early infants are able to represent cardinal as well as ordinal numerical information is reported (Chapter 2). Finally, the two predominant models of the development and structure of numerical knowledge in the first year and months of life are described (Chapter 3). Both these models posit the existence of an inborn system implied in numerical performance: An object-tracking system for object representation, called Object-file system (e.g., Carey, 1998; Uller, et al., 1999), and a numerical estimation system, called Analog magnitude system (e.g., Dehaene & Changeux, 1993). Recently, a new theory (i.e., ATOM model) has extended the application of analog magnitude system to continuous as well as numerical quantities, based on ordinal knowledge (Walsh, 2003). Whereas the Object-file system detects and represents objects with their spatio-temporal features and only subsequently extracts numerosity information; Analog magnitude system is specifically implied in the elaboration of quantity information, and its application is modulated by the numerical ratio of the values compared (i.e., ratio effects).
Strikingly, despite the assumption that both of these systems are innate, no studies have explored if the specific abilities required for object representation by Object-file system (e.g., Simon, 1999) and for quantity elaboration by Analog magnitude system (e.g., Gelman & Gallistel, 1978; Walsh, 2003) are present in early life (but see, Antell & Keating, 1983).
Starting from this lack, I tried to investigate the origin of numerical knowledge running three studies aimed at two main goals. The first aim has been to verify the presence of the abilities required by the Object-file system at birth. Specifically, Study 1 (Chapter 4) is aimed to investigate whether in the first days of life the capacity to form categorical representations of spatial relationships between visual objects is present. Using the visual habituation or familiarization technique five different experiments have demonstrated that newborns not only are able to discriminate an object's position with respect to a landmark (Experiment 1, 2, and 5), but they are also able to recognize a perceptual invariance between left/right spatial relations in condition of low (Experiment 3) and high-perceptual variability (Experiment 4) of the object. Altogether, evidence from Study 1 reveals that from birth, infants are able to treat spatial relationship between objects in a categorical manner, at least when the spatial relationship involves only two objects and they are easily discriminable from each other.
Study 2 (Chapter 5) supports and extends these results showing that newborns are able to detect spatial ordered sequences of three objects, arranged in accordance with left/right spatial-relation principles. Specifically, using the familiarization technique, Experiments 6 and 7 have shown that 3-day-old infants are able to detect a spatial order sequence based on the spatial relations between three elements at least when temporal information are not available.
Altogether, the findings obtained in Study 1 and 2 suggest that from birth human cognitive system detects, processes and represents object based on their visuo-spatial features supporting the hypothesis of the existence of an early general and automatic attentive system of object tracking based on visuo-spatial processing (e.g., Scholl & Leslie, 1999; Simon, 1997; Uller, et al., 1999).
The second main goal of the present research is to investigate the presence, in the first months of life, of ordinal representational competencies implied by a general representational system of magnitude (i.e., ATOM; Walsh, 2003). To this end, the presence and the features of a representational system of ordinal magnitudes at birth and at 3 months of life have been tested.
Specifically, Study 3 (Chapter 6) has ascertained when the ability to detect ordered sequence of continuous magnitudes arises. Collected data have demonstrated that 3-month-old infants (Experiment 9) were able to discriminate between a monotonic continuous magnitude ordinal sequence (e.g., going from the smallest magnitudes to the largest) and a non-monotonic sequence (i.e., random order). Conversely, newborns did not show this ability (see Experiment 8), even if they have showed the ability to recognize a spatial ordinal magnitude sequence when they are required to discriminate a monotonic from a non-monotonic sequence, in conditions of lower perceptual variability (Experiment 10).
Altogether, the data from Study 3 have demonstrated the abilities of detecting ordinal continuous magnitude arise even in the first months of life, earlier than evidenced in literature (Brannon, 2002). Moreover, these data allow outlining the developmental trend of ordinal knowledge in the first months of life, highlighting that ordinal competences develop in the first 3 months of life, at least for continuous magnitudes.
In conclusion, these data seem suggest that, at birth, the human cognitive system elaborates objects, whereas the ability to elaborate quantities appears very fragile. Consequently if the capacity for objects processing develops before quantity processing, it is arguable that from the first days of life, infants posses the abilities necessary to elaborate small numerosities, yielded by an Object-file system, and that subsequently the ability to elaborate analog magnitudes develops.