ICIS Conference - Brighton 2000
Tuesday 13:30 to 15:20 Main Hall
Knowledge of causality, intentions, and desires
Details of individual items:
Pre-schoolers imitating failed attempts: reading intentions vs. object affordances
Tony Charman, Chi-Tai Huang
A modification of Meltzoff's (1995) experimental design was used to testimitative responses to simple actions on a series of objects. Our aim wasto question Meltzoff's conclusion that when 18-month-old infants' imitate afailed attempt they are necessarily reading and reproducing the adult'sintended acts. Meltzoff used three conditions where the infant saw eitherfull demonstrations of the actions, a 'failed attempt' to produce theactions [e.g. the experimenter's fingers slipped off a pair of dumbbells asthey attempted to pull them apart], or a control condition in which theadult manipulates the object in some other way. We added a novel 'objectaffordance' condition, where only the end state of each of the target actswas seen by the infants, with the movement taking place behind a barrier.We also recorded the child's first two discrete actions on the objects,since we felt that one way to help us understand the goal of the infant wasto examine the sequence of actions produced as well as whether or not theyproduce a particular response within the response period. The results weresimilar to Meltzoff's findings but critically children watching the failedattempts did not produce the target acts more than they did in the objectaffordance condition. However, in this condition children tended to producethe target acts at their second action, preceded by other (exploratory?)acts as their first action.At least three aspects are relevant in the imitation situation. The actionsthat a particular objects affords, the action produced by the experimenterand seen by the child, and the underlying intentional stance whichmotivates the experimenter to act on the object in the manner in which theydo (whether or not they are successful in enacting this intention).Meltzoff's interpretation of infants' production of the target acts in thefailed attempt condition may conflate these three aspects. We will outlinefurther modifications to the paradigm that we are currently piloting inorder to help us better understand how object affordance, the reproductionof motor schema, and reading intentions, contribute to the imitationsituation.
8- to 10-month-olds perception of causation-at-a-distance
Anne Schlottmann, Luca Surian, Sarah Hesketh
This study replicates and extends previous work on infant sensitivity to causation at-a distance in reaction events (Schlottmann & Surian, in press), an ability that could support developing social cognition. Our approach is analogous to the one that previously established infant sensitivity to contact causality in launch events (Leslie & Keeble, 1987). 8- and 10- months-olds were habituated to computer animations involving two rectangles moving either rigidly or nonrigidly (both had the same translation speed; adults consider the nonrigid elongation-contraction animal-like.) Red started stationary on the left, blue in the middle. Red then moved towards blue, stopping before reaching blue's initial position. Infants in the 'reaction' group saw blue move in turn, starting before red stopped. The 'pause' group, in contrast, saw blue move after red stopped. For reaction, but not pause events, adults judge that blue moves to avoid capture, an impression enhanced by animal-like agents. Such reports of intentional causality in events without spatial contact between the agents complement adults' reports of physical causality in contact events. After habituation, infants saw a familiar and a reversal test. If during habituation and familiar test red moved towards blue (left to right), on reversal blue moved towards red (right to left). Reversal involved identical changes in spatiotemporal direction for all groups. However, causally interpreted events additionally involve a change in roles (the 'hunter' becomes 'hunted'). If infants perceive some role reversal, dishabituation should be enhanced compared to noncausal events. Our previous study found this -- with agents moving nonrigidly. The present study also asks how this kinetic feature contributes to the perception. Results (for 37 8- and 33 10-months-olds so far) show more recovery from familiar to reversal test in the reaction (10.5 to 19.8 secs) than the pause group (13.7 to 17.8 secs) when the agents moved nonrigidly. This replicates our previous finding. When the agents moved rigidly, recovery was less, but the same pattern appeared: In the reaction group, looking time increased from 11.5 secs on the familiar test to 15.4 secs on reversal, but in the pause group it only increased from 10.8 to 12.4 secs. The group x trial interaction, reflecting more recovery in reaction than pause events, reaches F(1,62) 3.05, p 0.086 so far. The shape x trial interaction, reflecting more recovery with nonrigid than rigid agents, reaches F(1,62) 3.39, p .070, but the 3-way interaction has F < 1, i.e., the two effects appear independent (in ANOVA on log looking times to normalise variances). No age differences were apparent in these recovery patterns (Fs < 1). This study again shows infants' sensitivity to causality in simple sequences in which two agents appear to interact at a distance. It also suggests their perception might be affected by the type of agent involved. Whether infants enhanced interest in reversal of nonrigid motion is due to low level features, or an emergent capacity to link some kinetic stimuli with 'animacy' remains an open question.Acknowledgement: This study was supported by the ESRC. References: Leslie, A. M., & Keeble, S. (1987). Do six-month-old infants perceive causality? Cognition, 25, 265-288. Schlottmann, A. & Surian, L. (in press). Do 9-months-olds perceive causation-at-a-distance?
The Ôanimal-ball experiment': a new paradigm to study causal attributions of animate motion in infancy
Sabina M. Pauen
It has recently been suggested that 7-month-olds expect animate beings to start moving on their own, whereas they expect inanimate objects to remain stationary unless being set in motion by an external agent (Spelke, Phillips & Woodward, 1995). To test this hypothesis, a new paradigm has been developed: Infants saw a toy animal (a stuffed animal with a smiling face, a worm-like furry body, and leg-like appendages) and an inanimate object (a bright orange-blue plastic ball). The ball contained a hidden battery-driven motor, causing it to move apparently on its own. Infants saw 3 separate scenes involving these two objects. Each scene was presented for 15 seconds. In scene 1 as well as in scene 3, both objects were lying motionless at some distance from each other. In scene 2, the head of the toy animal (i.e., the ãnose') was attached to the ball. The experimenter set the ball in motion before the scene was revealed to the child. Hence, the ball was pulling the animal, and both were rolling around together. A pilot study with N 3- to 4-year-olds indicated that preschoolers perceive this event as an animal playing with a ball. It was hypothesized that infants make a similar causal attribution, perceiving the animal as the agent and the ball as the recipient of the action. In this case, they should show an increase in looking time for the animal from scene 1 to scene 3, and a decrease in looking time for the ball, because they expect the animal, rather than the ball to start moving again during scene 3. An analysis of variance with object (ball, animal) as independent variable, and mean looking time during scene 1 and scene 3 as repeated measurement factor confirmed this hypothesis. We currently conduct several control studies to rule out alternative interpretations: These control studies include a condition with (a) the animal and the ball moving independently from each other, (b) only the ball, (c) only the animal or (d) no object moving during scene 2. Results of the whole set of experiments will be reported and implications for infants knowledge about animate motion will be discussed.
Pre-schoolers' access to explicit understanding of causalities
Francois Lefebvre, Jacqueline Nadel
The purpose of the current study was to examine whether the development ofexplicit understanding of intentional causality relies on a domain-specificcognitive system distinguishable from cognitive capacity in other domainsof causal cognition. Three groups of 20 Parisian pre-school children (10 females and 10 males)aged 3.5, 4.5 and 5.5 were presented 24 causal comic strips. The experimental design consisted of 3 sets of 6 scripts. The first setdepicted events involving objects affected by a physical causality (forinstance, one tomato is progressively falling from a full basket oftomatoes, see figure 1a). The second set depicted events involving childrenaffected by a physical causality with objets (for instance, a baby crawlingon the floor is caught on the electric wire of a lamp, see figure 1b). Thethird set presented children acting as agents in order to achieve a goal(for instance, a young boy reading on an armchair looks bothered by a fly,see figure 1c). The presentation of the different scripts was randomised.Each comic strips consisted of a total of 6 cards: 3 cards telling thestory and three answer-cards among which one gives the correct end of thestory.After a training period with 6 scripts, the children were asked to selectone answer-card to complete each story. The children had to give verbaljustifications of their choice. The narratives of all completed scriptswere recorded.Results suggested that children under 4 were not able to propose a logicalend to any story-cartoon, whatever the causality in play. After 5.5,children were mostly at ceiling for the three kinds of causal reasoning.Between 4 and 5 of age, children tended to explain a person's involvementin a physical event as due to an intentional action.
Understanding the intention of others: re-enactment of intended acts or spontaneous discovery?
Alessandra Assanelli, Laura D'Odorico
An investigation of infants' understanding of others as psychologicalbeings having mentalstates can be carried out by means of different paradigms, some of whichare based on imitation.Meltzoff (1995), for example, introduced a fairly direct experimentalparadigm designedto address this issue. The author makes use of infants' tendency to imitatewhat they see todemonstrate that when confronted with an adult who tries, but fails, toperform a given act,they re-enact not what they actually see the adult do, but what she/heintended to do.In his study, Meltzoff used as test stimuli 5 objects which the child couldnot have seenor played with before, because they were specially constructed in thelaboratory.Comparison of 18 month old toddlers' behavior in four independent groups(two demonstration groups: target action and intention; two control groups:baseline and adult manipulation) showed that children in the intentiongroup performed similarly to children in the target group anddifferently from children in the control groups.As infants had the experimental objects at their disposal for only twentyseconds alternativeexplanations of Meltzoff's results could be suggested: a) objectsthemselves have specific demand characteristics which induce the infant toproduce the target acts, independently of the adult demonstration and b)infants in the intention group are quicker to discover them because theexperimenter's failed acts focused their attention on the part of theobjects needed to producethe target action.Our research was designed to test this alternative interpretation byreplicating the original study, but:a) allowing infants to manipulate the experimental objects for 60-s periodsinstead of 20-s;b) comparing two age groups (12 and18 months).Moreover we analyzed data by means of log-linear analysis, assuming thatthe number of childrenperforming the target acts in the different groups is a more suitablemeasure to study developmental phenomena (Gustafson and Grenn, 1991)compared with the measure used byMeltzoff (number of target acts produced by infants).Our data which show no difference between the number of infants who producethe target action in intention and control groups, at both 12 and 18months, are consistent with our hypothesis that Meltzoff's results could beinterpreted alternatively in terms of 'velocity' of the response exhibitedby the Intention group compared with the Control group.ReferencesMeltzoff, A. N. (1995). Understanding the intentions of others:Re-enactment of intended acts by 18 month old children. Developmental Psychology, 31, 838-850.Gustafson, E. e Grenn, J. A. (1991). Developmental coordination of crysounds with visual regard and gestures. Infant Behavior and Development, 14, 51-57.
Self-awareness and other's intentionality-awareness at 18 months
Francesca Bellagamba, Luigia Camaioni
It is argued that the acquisition of self-awareness, which takes placearound the middle of the second year of life, not only re-organizesself-experience but is related also to the understanding of others'intentionality. The child becomes able to take into account the other'sperspective and to infer other's intentions even when they are notexplicitly manifested in action or when they differ from the child own'sintentions.The present study aims at investigating the relation between the ability tounderstand other's intentionality and self-awareness at 18 months of age.Self-awareness is defined as the ability to take the self as a symbolicobject that can be held in mind (and eventually shared with another person)independently from its direct perception. To evaluate self-awareness thefollowing tasks were used: a) passing the mirror self-recognition task(according to Asendorpf et al., 1996 revised version); b) referring to selfthrough pointing when the experimenter requests to localize oneself; c)referring to self through the own proper name or a first-person pronoun (band c adapted from Pipp-Siegel, 1997).The ability to understand other's intention was assessed through animitative task devised by Meltzoff (1995), which requires the child toinfer the adult's intended act by watching only some failed attempts toperform it.All the experimental tasks were administered to 30 Italian children (17girls; 13 boys) of 18 months of age in a laboratory playroom at theUniversity of Rome 'La Sapienza'. The Uzgiris-Hunt Scales (1975) and theParent Report for communicative and linguistic development in the secondyear of life (Camaioni et al. 1992), were administered to the children inorder to asses their cognitive and communicative abilities. Mothers werealso asked to fill a list of internal state words (adapted from Camaioni &Longobardi, 1996) when used by their children. The tasks evaluating self-awareness and other's intentionality-awarenesswere positively correlated. Children who were more able to infer other'sintentionality also get a higher score on the self-awareness tasks. Theassociation between the ability to infer other's intentionality and theability to take the self as a mental object at 18 months appears to begenuine, since the children examined were quite homogeneus as far as theircognitive abilities are concerned. However, those children wholinguistically referred to self and to other's intentionality, wereslightly more advanced than their age-mates in terms of communicativeabilities.
Do 3- to 9-month-old infants consider communicative intentions in others?
Infants start to take an intentional stance as they consider the motivesand intentions that guide others behavior and actions. The intentionalstance is thought to announce unique forms of human social cognition suchas imitative learning, language development, and symbolic play. Despitea general agreement about the role of the intentional stance in thedevelopment of human social cognition, debate remains regarding thedevelopmental origins of the intentional stance. At one end of thedebate some researchers argue that infants begin to take an intentionalstance at the end of the first year when they start to engage in triadicsocial behaviors such as social referencing, attention following, andcommunicative gesturing (e.g., Tomasello, 1995). On the end of thedebate, researchers suggest that infants start to understand otherscommunicative intentions in the early months of infancy (e.g., Murray &Trevarthen, 1985). The current study was designed to establish when young infants start toconsider others communicative intentions. Three-month-old (N 40),6-month-old (N 40), and 9-month-old infants (N 40) engaged in a 1minute normal interaction with a female experimenter for 1minute. Theywere then tested for their reaction to broken communicative contact (i.e.,still-face episode) for 1 minute. During the still-face episode, theexperimenter either broke contact with the infant to pose a neutral blankstill-face (Still-face Toward condition), to look at the infants motherwho entered the room while reading a book aloud (Interruption Condition),or to look away at a wall (Still-Face Away Condition). Infants affectivebehavior (e.g., gazing, smiling) and attempts to re-engage the socialpartner via social initiatives (e.g., touching, vocalizing, clapping) wereassessed for each minute of the study. Results indicated that infants at all ages engaged in less gazing towardthe experimenter during the still-face periods compared to the precedingnormal interaction. In addition, they manifested more vocalizing towardthe experimenter when she posed a still-face toward them (Still-faceToward Condition) compared to when she looked away (Still-Face AwayCondition or Interruption Condition). Compared to 3- and 6-month-olds,9-month-old infants manifested more overall social initiatives toward theexperimenter during the still-face episodes compared to the precedingnormal interaction. However, there was little evidence that infantsconsidered the motive underlying the experimenters behavior. Infants didnot differentially manifest social initiatives as a function of the reasonwhy the experimenter broke communicative contact. These findings suggestthat when social contact is abruptly disrupted, young infants do not takeinto consideration the communicative intentions that underlie a socialpartner's behavior.
Development of the ability to infer desires in 9- and 12-month-olds
Joanne Barna, Maria T. Legerstee
Recently, researchers have formulated theories on the developmentof a theory of mind (Baron-Cohen, 1994; Wellman, 1993). Infants have atheory of mind when they can make inferences about the mental states ofothers. Although infants are unable to attribute representational mental states to others such as beliefs, it has been suggested that they can attribute intentions to others based on observable states ofaffairs (Wellman, 1993). Wellman proposes that between 18 to 24 months, infants reliably demonstrate such ability. In contrast toWellman, Baron-Cohen argues that 9-month-old infants are capable ofimputing intentions to others. Few studies have been conducted with young infants that investigate their ability to infer desire in others,and that utilize nonverbal methods. Thus, we conducted a habituation study to determine whether 9- and 12-month-old infants can ascribe desire to others when the indicator of desire is positive affect. Inparticular, the infants were tested on whether they understood theprinciple of human behavior that people are more likely to reach for and hold objects that they desire. The results suggest that 9- and12-month-old infants were able to use emotional cues to infer desire inothers.