Perceptual and positional saliencies influence children’s sequence learning differently with age and instructions at test

ABSTRACT

There is growing evidence that, faced with a complex environment, participants subdivide the incoming information into small perceptual units, called chunks. Although statistical properties have been identified as playing a key role in chunking, we wanted to determine whether perceptual (repetitions) and positional (initial units) features might provide immediate guidance for the parsing of information into chunks. Children aged 5 and 8 years were exposed to sequences of 3, 4, or 5 colours. Sequence learning was assessed either through an explicit generation test (Experiment 1) or through a recognition test (Experiment 2). Experiment 1 showed that perceptual and positional saliencies benefited learning and that sensitivity to repetitions was age dependent and permitted the formation of longer chunks (trigrams) in the oldest children. Experiment 2 suggested that children became sensitive to perceptual and positional saliencies regardless of age and that the both types of saliencies supported the formation of longer chunks in the oldest children. The discussion focuses on the multiple factors intervening in sequence learning and their differential effects as a function of the instructions used at test to assess sequence learning.

KEYWORDS:

• Chunks
• Development
• Parsing
• Perceptual and positional saliencies
• Sequence learning

Acknowledgements

The authors are very grateful to Stéphane Argon, Patrick Bard, Laurent Bergerot, and Philippe Pfister, who designed and programmed the video game. We also thank Tim Pownall for his very careful correction of the English of the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. If chunking generally refers to a strategy that we intentionally use in daily situations for reducing cognitive demands in various tasks (to concatenate digits in a series of numbers to learn and remember a phone number, for instance), chunking is also considered as a mechanism of implicit learning. As claimed by Perruchet (2008, p. 608), “the fact that implicit learning leads to the formation of chunks is largely consensual”. However, there are two ways to consider chunking in the domain of implicit learning (see Perruchet & Pacton, 2006, for a discussion). First, chunks could be inferred from the statistical distribution of the material (e.g., Saffan & Wilson, 2008). Second, chunks could emerge without prior statistical computations. The initial formation of chunks could be guided by prior knowledge or sensitivity to salient features, for instance, but only chunks consistent with the statistical structure of the environment are consolidated and become relevant cognitive units (Perruchet & Vinter, 2002; Servan-Schreiber & Anderson, 1990). The present paper thus aimed to investigate the sources of information that contributed to the incidental formation of chunks.

2. The choice of the fifth colour, turquoise, may seem curious. However, preliminary investigations revealed that turquoise was a good candidate compared with other alternatives. Black and white colours were rejected because these colours within the sequences were interpreted as errors or omissions of colours by the children and thus elicited their attention in a problematic manner. Other secondary colours were judged less attractive than primary colours by children and potentially “under-selected” in consequence during the generation task. As a precaution, we limited the possibility to present blue and turquoise as consecutive/adjacent colours within the sequences, especially at the beginning of the flags. However, this case could not be totally excluded since we aimed to present each child with a different instantiation of the series to prevent any order effect of colour presentation.

3. The number of flags of three, four, and five colours was estimated in order to compose a 20-minute learning phase with some repetitions of each flag. The length of the flags was inspired by previous studies in adults and children using letter strings that contained generally two to five letters. However, three-colour strings were the minimal length for investigating competitive chunking between the first and second bigrams of the sequences. In addition, only two legal first bigrams and two legal second bigrams were authorized in the REP or in the non-REP sequences. So it was possible to generate only 2 three-colour strings for each condition, REP and non-REP, while more possibilities were offered in the production of the four- and five-colour strings.

4. An anonymous reviewer pointed out the possibility that the use of 25 coloured squares during the generation task might have disadvantaged the younger children, increasing visual search demands and the time needed to reproduce flags, thus imposing higher memory demands to young children during the generation task. For this reason, and others mentioned latter in the manuscript, Experiment 2 introduced a recognition task, less demanding in cognitive resources and therefore more age independent.

5. Readers may wonder whether this task can be truly considered as “incidental” since the attention of the participant is drawn to the flags by the following instructions: “the pandas will show you their pretty flags”. Because implicit learning was first defined as an automatic learning process, in opposition with controlled processes of explicit learning, “implicit” and “attention” are usually considered mutually exclusive terms. However, a growing body of evidence shows that selective attention is involved, and is necessary, during an implicit learning episode (e.g., Hoffmann & Sebald, 2005; Hsiao & Reber, 1998; Jiménez & Méndez, 1999; Pacton & Perruchet, 2008). The fact that the instructions referred to the flags was not a problem per se. The only characteristic to meet is that “implicit learning proceeds without participants’ intention to learn” (Perruchet, 2008).

6. The scoring procedure did not take positional constraints into account mainly because positional information may not be learned in the case of micro-rules learning, incomplete exemplars (possible in a brief exposition), or fragments/chunks learning. Position is therefore not a decisive argument to assess implicit learning of chunks, which is what we precisely investigated in our study. In addition, a fragmentarist view would predict that fragments in salient positions (at the beginning of the sequences, for instance), which intrinsically embed positional information, would be advantaged by a scoring procedure with positional constraints, while the fragments in non-salient positions (second bigrams for instance) would be disadvantaged (e.g., Perruchet, 1994), making unfair the evaluation of competitive chunking between these units. Though the scoring procedure did not consider positional constraints, sequences with positional constraints were, however, preferred because positional salience was one of the type of saliencies we aimed to study. It is indeed very likely that fixed positions favour the detection of regularities and facilitate implicit learning processes (e.g., Mathews et al., 1989).

Additional information

Funding

This research was supported by a grant from the Conseil Régional de Bourgogne Franche-Comté.