The knowledge of the preceding number reveals a mature understanding of the number sequence
Introduction
One of the fascinating questions in the development of early numerical skills is understanding how children learn the numerical meaning of number words and Arabic digits, a process known as symbol-grounding (Harnad, 1990; Leibovich & Ansari, 2016). Children learn to enumerate objects in their environment by respecting three counting principles: each item in a set must be associated with only one number word (one-to-one correspondence), the counting list must be recited in the established order (stable-order principle), and the last recited number word denotes the numerosity of the set (cardinality principle; Gelman & Gallistel, 1978). The development of counting skills is widely assessed using the Give-a-number task (GaN task), in which children are presented with a large set of items (e.g., 15 tokens) and asked to give different numerical quantities. Children sequentially learn the meaning of number words from one to four and correctly give them when requested. The limit of four corresponds to the number of objects children can keep in visual short-term memory (Feigenson, Dehaene, & Spelke, 2004; Knops, Piazza, Sengupta, Eger, & Melcher, 2014; Piazza, 2010; Piazza, Fumarola, Chinello, & Melcher, 2011). At this stage, children are defined as subset-knowers because their cardinal knowledge is still limited to the first number words (Le Corre & Carey, 2007; Sarnecka & Carey, 2008). Then, children crucially understand that next counted object corresponds to the next number word in the counting list (i.e., successor function; Carey, 2004). Children become cardinal-principle knowers (CP-knowers), and they can correctly count both small and large numerical sets. The acquisition of the cardinality principle represents a crucial milestone in the development of early numerical skills: CP-knowers have understood how counting represents numbers (Gelman & Gallistel, 1978; Sarnecka, 2015). An early mastering of the cardinality principle enables children to understand the numerical meaning of numbers and constitutes an advantage for acquiring later mathematical skills (Chu, vanMarle, Rouder, & Geary, 2018; Geary et al., 2018).
The present study further explores the symbolic numerical knowledge in preschool children who were (mainly) classified as CP-knowers. In particular, we examined preschoolers’ understanding of the exact numerical magnitude that is associated with number words and Arabic numbers and its relation with other numerical skills.
The mapping between numerical quantities and the counting list is usually assessed by asking individuals to estimate the number of items in briefly presented visual sets (Izard & Dehaene, 2008; Reeve, Reynolds, Humberstone, & Butterworth, 2012). The quick presentation of the target numerosities prevents the use of serial counting and allows to assess a mapping that is not the byproduct of a procedure. CP-knowers can precisely estimate small numerosities (<3-4 items), an effect called subitizing, which depends on the parallel individuation of distinct items in the set via the Object Tracking System (OTS; Kaufman et al., 1949; Le Corre, 2014; Odic, Le Corre, & Halberda, 2015; Revkin, Piazza, Izard, Cohen, & Dehaene, 2008; Sella, Lucangeli, & Zorzi, 2018; Trick & Pylyshyn, 1994). Large numerical quantities, instead, are processed via the Approximate Number System (ANS), in which each numerosity is represented as a Gaussian curve of activation whose width increases with numerical magnitude (Feigenson et al., 2004; Piazza, 2010). Only some CP-knowers, called mappers (i.e., CP-mappers), have established a linear mapping between the ANS and the counting list (i.e., ANS-to-word mapping) whereas some CP-knowers still lack such mapping (i.e., CP-non-mappers). Linearity is usually operationalised as a positive slope when regressing the estimates on target numerosities (Crollen, Castronovo, & Seron, 2011; Izard & Dehaene, 2008). The presence of a linear relation between estimates and target numerosities marks the understanding of the later-greater principle, that is, later number words in the counting list correspond to large numerical quantities (Le Corre & Carey, 2007; Le Corre, 2014). Children who master the cardinality principle should know that the late number words in the counting list represent large numerosities. However, this is not the case. Children acquire the later-greater principle after the cardinality principle (Le Corre & Carey, 2007). Accordingly, only CP-mappers provide estimates that increase with large numerical quantities whereas CP-non-mappers usually provide the same estimate (e.g., “five”) for large target numerosities (Odic et al., 2015). Crucially, CP-mappers have a better understanding of the symbolic system as indexed by their superior performance in comparing number words (Le Corre, 2014; but see Sella, Lucangeli, & Zorzi, 2018).
Other studies have suggested that the ANS-to-word mapping (and more generally the ANS) has a marginal role in the development of symbolic numerical skills (Carey, 2001; Lyons & Beilock, 2011; Lyons, Ansari, & Beilock, 2012; Lyons, Price, Vaessen, Blomert, & Ansari, 2014). Conversely, number words and Arabic digits acquire their numerical meaning when children learn their reciprocal relation, in a symbol-to-symbol association (Reynvoet & Sasanguie, 2016). In this vein, the knowledge and the experience in accessing the numerical sequence play a crucial role in the symbol-grounding. The numerical sequence can be represented in its verbal and visuospatial format, respectively, the counting list (i.e., “one”, “two”, “three”, “four”, …) and the number line (i.e., 1-2-3-4-…).
Children early memorise the counting list and can recite it forward. However, this does not imply that children have grasped the directional property of the counting list, that is, adding one item to a set leads to the next number word (i.e., n+1) whereas removing one leads to the previous number word (i.e., n-1). Sarnecka and Carey (2008) developed the unit task, whereby children watched the experimenter adding one or two items (n+1 and n+2) to a box already containing four or five items. Children had to say the number of items in the box after the manipulation by choosing between the n+1 and n+2 answer. Similarly, in the direction task, children were presented with two sets both containing five objects, then one object was moved to one set to the other and children were requested to indicate which set had now four or six objects. CP-knowers performed the unit and direction task above the chance level as they grasp the directional property of the counting list whereas subset-knowers displayed poor performance. However, not all CP-knowers could adequately perform the two tasks (see also Davidson, Eng, & Barner, 2012). Surprisingly, all CP-knowers can proficiently add up items to a set when performing the GaN task, but not all of them can perform the same adding-one transformation in the unit task. This discrepancy might emerge from the fact some CP-knowers are implementing a rote behaviour in the GaN task without really grasping the meaning of the manipulation. In this light, the implementation of a task capable of assessing the knowledge of the successive and preceding number across the counting list without requiring any motor routine would be a perfect tool to measure the knowledge of the directional property of the counting list in preschool children. Sella and colleagues (Sella, Lucangeli, Cohen Kadosh, & Zorzi, in press) showed children an opaque box and told them that inside there were some felt strawberries. Then, a strawberry was either added or removed from the box, and children had to tell the number of strawberries in the box after the manipulation. The starting number of strawberries in the box varied from two to eight to explore the ability to perform n+1 and n-1 transformations in the entire numerical interval from one to nine. CP-knowers displayed an accurate performance when one item was added (n+1) to the box as an index of their mastering of the successor knowledge. Conversely, their accuracy in performing the n-1 transformation was high with small starting numbers and decreased with large starting numbers. CP-knowers display an immature mastering of the predecessor knowledge, which is successfully applied to the first, but not the later, number words in the counting list. The performance in direction task correlated with the exact symbolic numerical knowledge, as measured using a number words comparison task, even when controlling for cardinality knowledge and memory capacity (Sella et al., in press).
The knowledge of the relation between numbers can also be assessed by measuring children’s familiarity with the number line, as done in the number-to-position task (Berteletti, Lucangeli, Piazza, Dehaene, & Zorzi, 2010; Siegler & Opfer, 2003). In the task, children mark the position of target numbers onto a visual line entailing a numerical interval, usually from 1 to 10 for young children. Some CP-knowers, called spatial-mappers, display a linear ordering of target digits on the visual line whereas other CP-knowers, called non-spatial-mappers, tend to place all the target digits in the middle of the line or at the start or end points (Sella, Berteletti, Lucangeli, & Zorzi, 2017). CP-spatial-mappers can reproduce the spatial arrangement of numbers and demonstrate better knowledge of the exact symbolic magnitude, as measured by a higher accuracy in a digits comparison task compared to CP-non-spatial-mappers (Sella et al., 2017). To further investigate the spatial mapping of numbers in preschool children, Sella and colleagues have developed a computerised task to separately assess direction, ordinality and accuracy of the spatial mapping (DOS task; Sella et al., 2018b). In the DOS task, a digit (e.g., 2) is presented in the middle of a horizontal line and the child is requested to place an additional digit (e.g., 1) in two possible locations, one on the left side and one the right side compared to the centred digit. The positioning of the target digit immediately determines the direction of the mapping, whether the child prefers to map numbers with increasing magnitude from left-to-right (e.g., 1–2) or right-to-left (e.g., 2-1). Then, another digit (e.g., 3) is presented, and the child is requested to place it. The positioning of the last digit reveals whether children respect the spatial order of numbers (e.g., 1-2-3) or not (e.g., 1-3-2) while also measuring the precision of the mapping, especially whether children respect the equidistance between digits (e.g., 1-2-3 vs 1-2---3). Preschool children mainly mapped numbers from left-to-right (according to the canonical direction in Western society), and the accuracy in ordering digits decreased with large triplets (e.g., 7-8-9) whereas the precision of mapping remained stable across triplets. Crucially, only spatial order related to the knowledge of the magnitude associated with digits, as measured in a digit comparison task (Sella et al., 2018b; Sella et al., in press). Children who can place digits in the correct spatial order within the number line can also indicate the exact numerical magnitude they represent.
Taken together, the results of the above-mentioned studies suggest that children’s exact symbolic numerical knowledge might emerge from the conceptual understanding of the directional property of the counting list and the knowledge of the spatial ordinal relation between numbers within the number line. The former represents the understanding that adding or removing items from a set is respectively associated with moving forward and backwards on the counting list. The latter indicates the knowledge of the position each number occupies within the number line. The counting list and the number line are both structures, whereby the ordinal position of a number can inform about its relative magnitude compared to the other numbers in the sequence.
In the present study, we explored the ANS-to-word mapping and the knowledge of number sequence and how these components relate to the understanding of the magnitude associated with number words and Arabic numbers. On the one hand, the cardinality principle and the ANS-to-word mapping represent a symbol-quantity relation, in which numerical quantities are mapped to symbolic numbers. On the other hand, the knowledge of the number sequence, as indexed by the directional property of the counting list and the spatial arrangement of numbers, reflect a symbol-symbol relation as numbers acquire a numerical meaning in relation to each other.
We selected a group of preschool children, who were more likely to be proficient counters (CP-knowers), to explore the knowledge of the symbolic system beyond the acquisition of the cardinality principle. We also ensured that all preschoolers in our sample knew the meaning of “numerically more” and could recite the counting list sufficiently high to perform the numerosity estimation task. Children performed a numerosity estimation task to assess the linearity of their ANS-to-word mapping. We used a direction task to measure the knowledge of the successive (n+1) and preceding number (n-1) in the counting list (Sella et al., in press). Children were presented with a box containing a starting number of tennis table balls, then one ball was added or removed, and children had to indicate the number of balls inside the box after the transformation. Crucially, this task does not require the implementation of rote behaviour, as the serial counting in the Give-a-number task. Moreover, we used both small and large starting numerosities, which could differentiate the knowledge of successive and preceding numbers within and outside the limit of the working memory capacity (i.e, subitizing limit in the OTS). When the box only contains 2 balls, children could perform the n+1 and n-1 transformation by tracking the number of items whereas, children had to rely on their knowledge of the directional property when the box contained a larger number of balls. We assessed the knowledge of the number line using a visual order task, whereby children saw three squares, two of them containing consecutive numbers: one in the central and one in the right (e.g., [], [2], [3]) or left square (e.g., [1], [2], []). Children had to say the number that goes in the empty square. Compared to previously used tasks, the visual order task resembles the transformations of direction task by separately assessing what number comes after (+1) and before (-1) in the number line. Moreover, it does not require any fine motor skills, as in the number-to-position and DOS task, and the effect of lexical knowledge was removed by reading all the numbers to children. Children chose the larger between two number words and Arabic numbers to assess their knowledge of the numerical magnitude of numbers. The direct request to compare the numerical magnitude of symbols and the fact that the response cannot be achieved by implementing a behavioural routine, as in the GaN task, make the comparison tasks the perfect tools to evaluate the knowledge of the exact magnitude associated with numerical symbols. In the number comparison task, error rates and reaction times increase as a function of the numerical distance (i.e., distance effect) and the size of the numbers (i.e., size effect; Moyer & Landauer, 1967). These effects relate to a general ratio effect as the representation of numbers overlaps with increasing numerical magnitude, as in the ANS, which is supposed to be the underlying mechanism of both non-symbolic and symbolic number comparison (Dehaene, 1992; Piazza, 2010), even though alternative models for number comparison performance have been proposed (Krajcsi, Lengyel, & Kojouharova, 2018; Verguts, Fias, & Stevens, 2005; Zorzi & Butterworth, 1999). Moreover, the speed in comparing Arabic digits has been repeatedly associated with arithmetic skills (see Schneider et al., 2016, for a review) and has been found reduced in children with developmental dyscalculia (Rousselle & Noël, 2007), which makes the exploration of the same ability in preschool children relevant, even though based on accuracy rather than response time.
Most of the studies have focused the investigation of early numerical skills on the range from 1 to 10 in preschool children. Nevertheless, it should be considered that preschool children, especially CP-knowers, should in principle correctly associate numerical quantities with the corresponding number words in their counting list, which extends above ten. In this light, preschool children might already have preliminary knowledge of the numerical magnitude represented by numbers above ten (e.g., Gilmore, McCarthy, & Spelke, 2007). Therefore, we chose to present numbers below and above ten to have a better description of children’s numerical knowledge of the broader numerical sequence. We used the same target numerosities and numbers across tasks to ensure perfect comparability of the performances. Finally, we assessed verbal and visuospatial memory as domain-general control measures.
We expect some of the children in our sample to display a linear mapping between the ANS and the counting list (Le Corre & Carey, 2007; Le Corre, 2014; Sella et al., 2018a). Children should display better performance in the n+1 trials of the direction task compared to the n-1 as children are more familiar with the process of adding up items as in the GaN task compared to remove items from a set (Sella et al., in press). We expect the knowledge of the spatial arrangement of numbers to decrease with large triplets (Sella et al., 2018b). Finally, we specifically examine the relative contribution of the ANS-to-word mapping, the knowledge of the counting list and the number line on number comparison skills. Such direct contrast should inform whether the knowledge of the magnitude associated with number words and Arabic numbers relates to children’s knowledge of the numerical sequence or the ANS-to-word mapping.
Section snippets
Participants
One hundred and four preschool children from five schools located in northeastern Italy took part in the study after parents, or legal guardians gave their informed consent. Parents or legal guardians also filled in a questionnaire regarding the family background and demographic information (e.g., nationality, parents’ education). We excluded those children who: failed to correctly choose the larger set in the dots comparison task at least in ten out of twelve trials (n = 5); failed to enumerate
Results
We ran the statistical analyses using the free software R (R Core Team, 2016) along with the BayesFactor package (Morey & Rouder, 2015) using default priors for Bayesian analyses. We reported Bayes factors (BF10) expressing the probability of the data given H1 relative to H0 (i.e., values larger than 1 are in favour of H1 whereas values smaller than 1 are in favour of H0; Wagenmakers et al., 2017, 2016). When comparing regression models, we reported the Bayes factors (BF) as the ratio of BFs10
Discussion
A proficient counting represents the first step in a long process that leads children to a mature understanding of the numerical meaning of number words and Arabic numbers. In the present study, we explored how preschool children performed a variety of tasks, which specifically assessed the mastering of different numerical concepts beyond the cardinality principle. To this aim, we ensured that all children involved in the study could proficiently count, knew sufficiently well the numerical
Conclusions
Despite the mastering of the cardinality principle, only some of the children in our sample have established a mapping between the ANS and the counting list. The direction task revealed a discrepancy between the mastering of the successor and predecessor knowledge. Children frequently said the successive number in the counting list even when one item was removed. The tendency to name the successive number might have inflated children’s mastering of the successor knowledge and, more generally,
Acknowledgements
The authors wish to thank the children and their parents for participating in the present study, as well as Giulia Alfiero and Sara Zennaro for their help in collecting data.
References (73)
- et al.
Representation of numerical and non-numerical order in children
Cognition
(2012) - et al.
Do analog number representations underlie the meanings of young children’s verbal numerals?
Cognition
(2017) - et al.
To infinity and beyond: Children generalize the successor function to all possible numbers years after learning to count
Cognitive Psychology
(2017) - et al.
Children’s early understanding of number predicts their later problem-solving sophistication in addition
Journal of Experimental Child Psychology
(2018) - et al.
Skills underlying mathematics: The role of executive function in the development of mathematics proficiency
Trends in Neuroscience and Education
(2014) - et al.
Does learning to count involve a semantic induction?
Cognition
(2012) Varieties of numerical abilities
Cognition
(1992)- et al.
Core systems of number
Trends in Cognitive Sciences
(2004) The symbol grounding problem
Physica
(1990)- et al.
Calibrating the mental number line
Cognition
(2008)
One, two, three, four, nothing more: An investigation of the conceptual sources of the verbal counting principles
Cognition
Numerical ordering ability mediates the relation between number-sense and arithmetic competence
Cognition
Children’s mappings between number words and the approximate number system
Cognition
Neurocognitive start-up tools for symbolic number representations
Trends in Cognitive Sciences
Subitizing reflects visuo-spatial object individuation capacity
Cognition
Basic numerical skills in children with mathematics learning disabilities: A comparison of symbolic vs non-symbolic number magnitude processing
Cognition
How counting represents number: What children must learn and when they learn it
Cognition
Unpacking symbolic number comparison and its relation with arithmetic in adults
Cognition
Preschool children use space, rather than counting, to infer the numerical magnitude of digits: Evidence for a spatial mapping principle
Cognition
Children’s understanding of counting
Cognition
Working memory deficits in developmental dyscalculia: The importance of serial order
Child Neuropsychology: A Journal on Normal and Abnormal Development in Childhood and Adolescence
Serial order working memory and numerical ordinal processing share common processes and predict arithmetic abilities
The British Journal of Developmental Psychology
Numerical symbols as explanations of human perceptual experience
Numerical estimation in preschoolers
Developmental Psychology
BVN 5-11: Batteria di valutazione neuropsicologica per l’età evolutiva
Executive functioning as a predictor of children’s mathematics ability: Inhibition, switching, and working memory
Developmental Neuropsychology
Cognitive foundations of arithmetic: Evolution and ontogenisis
Mind & Language
Bootstrapping & the origin of concepts
Daedalus
Ontogenetic origins of human integer representations
Trends in Cognitive Sciences
Under- and over-estimation: a bi-directional mapping process between symbolic and non-symbolic representations of number?
Experimental Psychology
Executive Function
Annual Review of Psychology
The representations underlying infants’ choice of more: Object files versus analog magnitudes
Psychological Science: A Journal of the American Psychological Society / APS
Split-half reliability
Early conceptual understanding of cardinality predicts superior school-entry number-system knowledge
Psychological Science
Sensory-integration system rather than approximate number system underlies numerosity processing: A critical review
Acta Psychologica
Cited by (24)
A refined description of initial symbolic number acquisition
2023, Cognitive DevelopmentCitation Excerpt :While several studies argue that CP-knowers show a qualitatively different knowledge of number properties compared to subset-knowers (Condry & Spelke, 2008; Lipton & Spelke, 2006; Sarnecka & Carey, 2008; Wynn, 1990, 1992b), others argue that CP-knowers perform more poorly than what would be expected if they had an appropriate conceptual understanding of numbers. For example, it has been demonstrated that some CP-knowers cannot estimate the numerosity of a visual set (Le Corre & Carey, 2007), cannot map symbolic numbers onto a spatial number line (Sella et al., 2017), cannot compare symbolic numbers (Davidson et al., 2012; Le Corre, 2014; Sella et al., 2017), and do not know how a set is changed when an item is added or removed (Davidson et al., 2012; Sella & Lucangeli, 2020). Overall, while validational viewpoints may help to determine whether children are subset-knowers or CP-knowers, the results are not straightforward, and there is no consensus in the literature as to which task (including various aspects of the GaN task itself) is optimal for appropriate validation.
The approximate number system cannot be the leading factor in the acquisition of the first symbolic numbers
2023, Cognitive DevelopmentCitation Excerpt :It is also assumed that CP-knowers understand many general properties of symbolic numbers (e.g., see Lipton & Spelke, 2006). Note that there are important debates on whether this knowledge generalizes to all numerical tasks (Davidson et al., 2012; Le Corre & Carey, 2007; Sella & Lucangeli, 2020) or whether the GaN task in its current form is the appropriate tool to measure this knowledge (Sella et al., 2021). A key question concerning this developmental pattern is what representation may be responsible for the observed pattern.
Core foundations of early mathematics: refining the number sense framework
2022, Current Opinion in Behavioral SciencesCitation Excerpt :Number relations involves knowledge of the relations between and among whole numbers (e.g. determining which quantity is more, less, or equal to another quantity) [10]. By engaging with number relations, children form preliminary understandings that numbers in the count list are represented linearly, with each number being exactly one more than the previous one [11,12], allowing them to order numbers and estimate their placement on a number line. Facility with number operations involves knowledge of how whole numbers can be taken apart and put together.
Numerosity sense correlates with fluent mathematical abilities
2022, Acta PsychologicaCitation Excerpt :Indeed, researchers have found that skill at digit comparison and parity judgment correlates with processing speed (Tavakoli, 2016; Zhang et al., 2016). Numerosity estimation is defined as the mapping between symbolic and nonsymbolic quantities without counting (Sella & Lucangeli, 2020). Although numerosity estimation is also a simple application of existing knowledge (quantity), it is a slow process because participants must constantly estimate the quantity and change it into symbolic numbers without counting (Bartelet et al., 2014; Beran et al., 2006).
The effects of preschool and age on children's early number skills
2022, Cognitive DevelopmentCitation Excerpt :The effect of one year of schooling on children’s number order again suggests that it might be improved via preschool education, which aligns with previous intervention work in this area (Clements et al., 2011). The observation that pre-schoolers’ knowledge of number order is critical to and predictive of their understanding of the meaning of number and to their subsequent mathematical development (e.g., Sella et al., 2020; Sella & Lucangeli, 2020) further motivates the focus on number order in early childhood education. The current study also found effects of schooling and age on children numeral identification.