Linguistically modulated perception: how words catalyze categorization

In this study subjects had to remember items in different conditions: the items were previously marked either by one word or by two words. It turned out that the presence of two different words made subjects less sensitive to individual distinctions between items and consequently resulted in poorer memory of such individual examples. The effect in the one word condition was the opposite. That effect – points to a spontaneous activation of the categorization process in the mere presence of the word and for the first time it was found on artificial material.

Role of sign in concept formation

In our previous studies, we found that in a situation where the formation of concepts takes place in conditions that hamper the selection of rule (e.g., the presence of «noise» objects), the sign facilitates learning. These results are consistent with other studies that show the role of the sign in strengthening categorization (Lupyan, Rakison & McClelland, 2007), selecting information about an object for the formation of meaning (Bloom, 2002) and helping children to transfer meanings (Gelman & Coley, 1990). In our current studies we are testing the hypothesis of sign polyfunctionality — the possibility that the sign could assume different functions (remembering, planning, monitoring, etc.) in different contexts and at different ages. We also explore on which properties of the sign depends its ability to perform these functions, most of which are related to regulation of the process of creating generalizations. We believe that not every object can act as a sign, but only one that differs from other natural properties of the object of categorization. We tested this hypothesis by creating tasks in which we change the properties of the sign, along with changes in the natural properties of the object of categorization. We are also interested in the effect on the concept formation of integrated sign consisting of several parts and sending in its form to different cultural symbol systems. So currently we test the hypothesis that the addition of a numeral to a category name would affect the categorization in children between the ages of 4 to 5 years.

Development of word meanings

Another area of our research is the ontogenetic development of speech and word meaning acquisition. We explore various influences on word meaning acquisition, such as fast mapping and shape bias, in two- to five-year-old children. Our previous research has shown the high value of such environment characteristic as dynamics of object’s features. We have shown that adding the dynamics of shape, color or other features of the object into the task makes the property of this dynamics more significant for categorization than static features.

Another research topic related to speech development is the influence of adults’ behaviour on how the child acquires the meanings of new words. At the present time we are testing the hypothesis that after observation adults’ actions with objects and perception of their speech children can learn both an association between name and object and a word meaning as a part of convention. Our data allow us to suppose that an invitation to the convention about an object is the function of getting child’s attention to the act of naming («Look! This is a <label>!»): the children performed mutual-exclusivity test only after presenting new label with the getting of attention to the naming. But the association between name and object can be learned without this: the children performed object-matching test even new label was presented in passing joint action with the object (BCCCD14, Implicit Learning 2014; in prep).

The selected question, which is interesting for our group, is how children idetntify words that are conventional. One of our experiments shows that the connection between objects during their demonstration and words pronouncing is importatnt cue of conventionality of words.

Another question we have study for is what is the basis of the expectation of novel word generalization in the beginning of word learning? There are many studies showing that the shape of the object is likely to be such a basis: 2-2.5 year olds tend to expect that the new word refers to a test object with the same shape (shape bias), but only if it is an artifact and not a substance or an animal (Booth&Waxman, 2002; Gathercole, 1997; Soja, 1992; Smith, 2000; Yoshida&Smith, 2001, 2003).
However, another approach considers that infants take the label as a symbol for a concept (Waxman & Gelman, 2009). Accordingly, the grounds for generalization must be conceptual rather than perceptual (Diesendruck, Markson, and Bloom, 2003; Yin, Csibra, 2015). We supposed that the basis for generalization is an adult’s demonstrated action with an object. To verify this supposition we conduct a “looking while listening” studies testing which similarity of the test object to the named object will be more important for 17-19-month-old participants: similarity in shape or similarity in the action performed with the object (coolaborative project with G.Csibra and A. M.Kovacs (CEU).

Mechanism of Overimitation Effect

The overimitation effect is a uniquely human phenomenon, in which children imitate irrelevant actions demonstrated by an adult with a new object. A contradiction exists between two explanatory principles in studies of the overimitation effect and imitation in general. On the one hand, children, following an adult’s demonstration, most likely ignore the rationality principle while imitating irrelevant actions. On the other hand, existing data demonstrate that children do not imitate actions that don’t lead to any outcome (Kiraly, Csibra, & Gergely, 2013).

Our studies tested the hypothesis that overimitation reflects children’s acquisition of “shared action” with an adult. First, we propose that within such acquisition children apply the rationality principle with the only purpose: to detect the action itself. Secondly, guided by the communicative situation, children identify the demonstrated action with the essence of the object. Analogously to the normative account, we suggest that children perceive the demonstrated action as a social norm in that case. Shared action acquisition does not assume understanding of its conditional character, as norm acquisition (in the ordinary sense) assumes.

Our experiments reveal that

  • children overimitate after confident and aware demonstration and do not overimitate after adult’s acting as if she has never dealt with the object before (Kotova & Preobrajenskaya, 2009);
  • сhildren followed a new set of irrelevant actions with the same object, if the adult varied them later (Kotova, 2013);
  • even if the infants observed the adult’s action resulting in interesting outcome (and this action was accidental), the infants prefer copying adult’s intentional action (in spite of the absence of an attractive result after it) (Kotova, Yudina, & Kotov, 2014);
  • children refrain from reproducing irrelevant actionsafter modification of the joint action, but not after another interaction change (in prep);
  • children refrain from reproducing irrelevant actionsafter modification of the perceptive characteristics of object, but not after causal structure of object change (in prep.).

Our current study devoted to search social cues that have an influence on overimitation in different ages, because we suggest that there is an “evolution” of these cues on the course of development. We have got the data that a verbal instruction can stop copying of irrelevant actions in 7-year-olds, and a picture of object without parts for irrelevant actions can stop it in 4-year-olds (in prep).

Social learning in emotional development

Studies concerning relation between social referencing and attachment are focused mostly on the matter of trust between adult and child (Dickstein, Thompson, Estes, Malkin, Lamb, 1984; Bradshow, Goldsmith, Campos, 1987), but they do not apply to the referential interpretation of the displayed emotion. Following Gergely and Unoka (2008), we consider emotion representation’s formation as consequence of child’s referential attitude towards emotions manifested by adults. This supposes that development of emotional representation relates to attachment security.

In our study, we do bring attachment security into correlation with tasks on social-referencing judgment varying emotional context. We can see different forms of emotion reaction to an object in securely and insecurely attached children: critically accepted, retainable out of the specific situation, and situationaly accepted, unstable without external cues, respectively. Our results confirm that adult’s emotion can be accepted as substantial cultural object feature, as like its label or function (Gergely, Csibra, 2006), but only in insecurely attached children. It is conductive to joint analysis in crosscultural studies of emotion (Fischer, Manstead, 2008; Diener, et al., 2002) and social learning (BCCCD2015; in prep).

We have found such correlation between prosocial behavior (i.e. instrumental helping) and attachment style. That is why we are testing some suppositions about the involvement of the social learning and the mutual understanding in prosocial behavior formation.

To test the hypothesis that at the end of the first year of life the ability to represent emotions (that is, ability to make inferences about a person’s behavior toward object after observing person’s demonstrated emotional evaluation of that object) is formed , we conduct the studies in the “expectation violation” paradigm, analyzing whether 15-18 month-old participants would expect a person to choose from two objects the one toward which she showed positive rather than negative emotions; and that the object toward which she showed tenderness would be stroked, in as opposed to an object in which the person has demonstrated an interest-in which case the expectation is that the person will manipulate by exploring that object (coolaborative project with I.Kiraly (Eötvös Lorand University; CEU).