Objectives To develop a child-friendly methodology to study children s representations of social support available from their personal relationships; and to examine children s representations of support from their pets compared to support from human relationships.

Design Participants were 22 year-3 primary school children aged 7-8 years. They were asked to list all the people and animals important to them and then to select a top 10 of most special relationships. Using a story-based methodology, children were asked who from their top 10 they would turn to if they were the child in the story.

Results Consistency in the data indicated that the children could reliably discriminate between different relationships in terms of the support functions they serve. Pets were often ranked higher than certain kinds of human relationship, and they featured prominently as providers of comfort, esteem support and confidantes for a secret. Confidence in these findings is gained through pets not being nominated for functions they could not realistically perform.

Keywords: children, pets, social relationships, social support

Pets are popularly thought to afford special benefits and enjoyment to children, which may explain why pet ownership is most frequent in families with children (Rehm 1993). These presumed benefits include the learning of responsibility for care, empathy for another living thing, and education in matters of illness, death and reproduction (Bryant 1990; Davis & Juhasz 1995; Melson 1998). Pets are also widely believed to provide a special form of companionship for children, this being the most frequently stated reason for acquiring a pet for children (Endenburg 1995).

Research into children and their pets appears to be divided into two broad approaches: research investigating the characteristics of the child-pet relationship (e.g. Bryant 1990; Endenburg 1995); and research comparing child pet owners with child non-owners and their social development in such social skills as empathy and support seeking (Covert et al. 1985; Van Houte & Jarvis 1995). Less attention has been paid to the role of the pet in a child's wider social network, comparing it with the roles and functions served by other (human) relationships. If what goes on between child and pet is like a social relationship then, for the nature and function of this relationship to be understood, it needs to be investigated as part of the child's wider network of social relationships together with the perceptions the child has of his/her other relationships.

Traditionally, research into children's relationships has focused on particular types of relationships such as relationships with parents, siblings or friends. Typically, each type of relationship is seen as having a small number of core features such as security and a secure base in attachment, conflict and caregiving in sibling relationships and reciprocity in friendship. However, more recent work indicates that similarities between different types of relationship are as striking as their differences. Dunn (1993) provides evidence that various types of relationships have many features in common. For example, mutual warmth, expressed affect, conflict, shared activity, comfort seeking and so on may feature not only in parent-child relationships but also in child-sibling relationships, child-friend relationships and many others. The social-support functions of children's relationships further illustrate Dunn's point that many relational characteristics and functions may be present in a variety of different relationships.

To assess these functions, self-report questionnaires have been developed for use with older children. Furman and Buhrmester (1985) developed a questionnaire for children aged 11-13 years which assessed both perceived availability and quality of support over the a range of relationship functions. The results provide evidence that children can reliably distinguish between the characteristics and functions of different relationships in their social networks.

For younger children, the use of self-report questionnaires has obvious limitations. Reid et al. (1989) suggested that research on assessing support networks in younger children should use instruments that are interesting and enjoyable, interactive, make use of visual materials, and be personally relevant and commensurate with the child's cognitive and emotional understanding. Moreover, there should be scope for investigating both family and non-family relationships.

Reid et al. (1989) devised a methodology called 'My Family and Friends' for use with children between 6 and 12 years old. Children were asked to list their family members, relatives, friends and their teacher, the names of these being written on individual cards. The children were then asked to rank these in answer to verbal questions directed to assess the instrumental, emotional, and companionship support they provide. A card barometer with a moveable indicator was then manipulated by the children to indicate the level of satisfaction the children perceived from the relationships to these forms of support.

This methodology was found to be acceptable and understood by even the younger participants. It had acceptable test-retest reliability and demonstrated that children were able to identify support availability from specified relationships in their social networks and to give assessments on the satisfaction received. Reid et al.(1989) distinguished between 'specialist' and 'generalist' support providers. A generalist is a person who was viewed by the child as providing a range of support functions over a number of situations, such as mothers and fathers, whilst a specialist was a person who was regarded as being a provider of a particular type or types of support. The results confirm that even young children can and do distinguish between relationships in terms of the support they provide.

Research on children's social networks has concentrated on a limited range of types of relationships. Often, researchers identify or constrain which types of relationship are to be investigated rather than permitting the child to select which relationship they consider to be important, and few studies have considered the potential role of pets in children's social networks. Yet recent findings suggest that children may regard pets as important relationships. Bryant (1985) investigated social support networks in children between 7 and 10 years by taking them on a 'neighbourhood walk' designed to incorporate cues from everyday activity and experience into the interview schedule and so improve reliability and validity. She found that children spontaneously generated the names of their pets, or neighbours' pets, as members of their social networks and described them as special friends or confidants. The identification of these two relationship functions had high test-retest reliability with correlations of 0.99 and 0.92, respectively. In a later paper, although not comparing support functions of pets and human relationships, Bryant (1990) examined the nature and benefits that may accrue to the child from his/her relationship with a pet. She identified four main factors: mutuality, a form of close companionship; enduring affection; self-enhancing affection, probably through elevated self-esteem; and exclusivity.

The study reported here investigated young children's perception of support, using a methodology which permitted the children to nominate their own significant relationships. In so doing, the children frequently nominated pets, so we were able to explore children's perception of support available from pets. The methodology had a lot in common with that used by Reid et al. (1989), but it allowed the children to directly express comparisons between different relationships as sources of support

Participants were a class of 22 children aged between 7 and 8 years (year 3) attending a middle school in a small town in the English Midlands. Thirteen participants were boys and nine were girls. Eighteen children owned pets, four did not.

The basic testing procedure involved two stages, which were carried out on 2 consecutive days. The procedure was repeated 7 days later to assess reliability.

The first stage of the procedure was to obtain a view of each child's social network as they, the individual child, perceived it. This was conducted with the class as a whole. The children were asked to make a list of their friends and their families, including any pets they may own, their teachers and other people who may be important to them, such as childminders or family members who did not reside in their household. No number was specified as to the size of the list, and help was offered to children who wished to list a person but were unable to spell their name or relationship.

When each child had completed his/her list, they were asked to select the 'top 10' most special relationships. The children were then presented with a range of materials with which to make cardboard figures to represent the characters in their top 10 selection.

The second stage of the procedure took place the following day. Each child was interviewed separately in a room near to the classroom. The children were asked to show the experimenter their top 10 figures and to introduce them by name and by relationship.

The experimenter then told the child that she would tell a number of short stories. These stories were designed to elicit the child's responses to a need for support and to identify those relationships perceived by the child as best providing the type of support required. The stories examined needs for comfort, companionship or to feel better about themselves (esteem support). Additional stories investigated the presence of conflict in relationships. For each story the child was asked, if they were in a situation like the child character portrayed in the story, which of their top 10 relationships they would most like to help them. Then, if that person were not available, who would they most like, and so on until they had selected five relationships (except for the conflict stories when they were only asked for three choices). The sequence of choices was taken to be the child's rank order of preference among his/her top 10 relationships for the particular story. The decision not to ask the children to indicate the order of preference for all of the top 10 relationships was based on observations during pilot studies that whilst children found it easy to select the first four or five choices for most scenarios (except the conflict stories), they frequently remarked that, although the remaining relationships were special, they would not approach any of them for that function described in the story. This was even more marked in the conflict scenarios for which children commonly nominated fewer relationships.

Participants were interviewed individually and eight short stories were told. Each story was accompanied by a storyboard, comprising pictures and words, which helped the child focus on the main character and the type of need he/she required. The dialogue was conducted using the cardboard figures of the child's top 10 as props to help the child indicate a chosen relationship, and to indicate to the child when one of the characters was not available. The stories can be summarized as follows.

The subject was shown a storyboard depicting a small boy in bed. The subject was told that the boy was poorly and could not go out to play and would have to stay in bed until he was better. As he was feeling a bit fed up and miserable, he would like someone to come to see him and help cheer him up a bit.

The subject was then asked 'If you were ill in bed, and you felt like he (the boy in the story) felt, who would you most want to come and see you?' The subject then had to choose one of his/her figures to represent who he/she would most want to come to see them if they were ill in bed. When the subject selected his/her first choice, the experimenter then removed the figure and said, 'But what if X (selected character's name) wasn't there, who would you choose?' The subject then selected their next choice. The procedure was repeated until five characters had been selected and were recorded by the experimenter as ranks one to five in the subject's choices. The characters were then returned to the subject in readiness for the next story.

The second story also examined the need for comfort, this time in the context of having to cope with a potentially frightening event. The subject was shown a storyboard and told of a small boy who had to walk home from school through a long dark lane. The boy thought it was very creepy and would have liked someone to be with him to stop him feeling scared. The subject was again asked 'If that was you, and you had to go down a long dark lane, who would you most want to be with you?' The subject then had to select his/her first choice of person in their top 10 to accompany them to stop them feeling scared. This first choice was then removed and the experimenter asked, 'If X wasn't there, who would be the next best?' This continued until five choices had been made from the available 10.

The first of two esteem-related stories concerned a girl who had been selected to take a big part in a school play. It was a difficult part and she had to practice lots of songs, her words and actions. She makes lots of mistakes while she is practising at school and at home. This makes her feel silly and embarrassed. She knows she needs to practice to make sure she is ready for the play, but she only wants to practice in front of people who won't make her feel silly or embarrassed if she does make mistakes. The subject was asked who they would choose as being best to practice in front of, if they had to do what the girl in story does. The procedure of removing each choice, as before, was continued for all the remaining scenarios.

The second esteem-related story told of a girl who is normally very sensible and organized, but who had a 'bad day' when lots of things went wrong. She was late for school, forgot her pencil case so that the teacher was annoyed, forgot her lunch, lost things and then falls over on the way home from school and ruins her best shoes. Everyone says 'Oh, you silly girl!' all day long and she feels very bad about herself because this is so unlike her normal behaviour. She needs someone who knows she is not really silly and who would be able to reassure her that it was just one of those days. The subject was then asked who would be most likely to be able to make them feel that they were not silly really.

The first story on confiding examined the telling/sharing of a special secret. The subject was told a story of a young girl who found a magic door at the bottom of her garden. The door led into a wonderful magic garden. The girl wishes she had someone to share in the excitement. A friendly wizard suddenly appears and welcomes her, but also warns her that she may only share her secret with one person. If she told any more than one other person, she would never find her way back into the magic garden. The subject was then asked who they would share the secret with, if they were ever in that situation.

This story depicted a small boy who was being bullied by children he did not know. It was making him very miserable and presenting quite a problem to him. He knows that he has to tell someone who can help him. The subject was asked to imagine that they were in that situation, and who they thought would be most helpful to tell of their problem.

A miniature plastic dustbin was placed in front of the subject. He/she was told that although people/pets can be special, we all get annoyed with them at times. When someone is especially annoying we can put them in the bin! The subject was asked, 'Who annoys you so much that you would put them in the bin first?' The subject made his/her first selection. They were then asked 'Who would be next in the bin because they annoyed you almost as much?' This exercise was limited to three choices only since many children indicated that only a minority of their relationships warranted being put in the bin for annoying them.

A second conflict scenario was presented to examine the subjects' perception of who they thought found they themselves (the subject) annoying. Subjects were asked 'Who do you think finds you annoying? Who would be the first to want to put you in the bin?' The subject selected the character they thought would be most likely to put them in the bin for being annoying. Again, the process was repeated for the selection of three characters.

The whole procedure, except for he construction of the first network list, was repeated one week later to test the reliability of the subjects' responses. On retest, subjects were told that the exercise should not be regarded as a memory test and that they should choose their top 10 relationships from their larger lists as honestly as they could, even if this meant they differed from their first selection.

The mean number of relationships generated by the children for their larger first list of all the relationships considered important to them was 15 (range 9-23), so for the majority (18) of children there was scope for their top 10 to differ between the two occasions. Of these 18 participants, 12 had perfect agreement on their choice of top 10; the mean value of kappa coefficients of agreement was 0.89. These findings indicate good test-retest reliability for the selection of top 10 lists of relationships, which implies that the children did not make their selections merely on the basis of temporary preferences prevailing at the time of testing.

Eighteen participants owned pets and listed these on their larger overall list. Of these children, 17 included one of more pets in their top 10: nine participants had either a cat or a dog in their top 10; four had a 'small pet' (rabbit, hamster, guinea pig or budgerigar) in their top; four had a cat or dog and also a small pet to their top 10. The child who did not include his pet in his top 10 owned a guinea pig. Some participants who owned more than one pet included them on the overall list but selected only one for their top 10 selecting, for example, a budgie out of a budgie and a hamster; a rabbit from a rabbit and a dog; and a dog from a dog and a guinea pig. This willingness to select pets as significant relationships supports Bryant's (1985) claim that children may regard pets as special friends and companions within a child's social network.

To examine the reliability of rankings of relationships, for each of the eight stories rankings from the first session were correlated with rankings from the retest session. The correlations on the diagonal of Table I are the test-retest correlations for each story. The high level of these correlations indicates that children are relatively consistent across sessions in their choice of relationships.

Correlations between different stories (in the off-diagonal cells in Table I) were generally lower than the test-retest correlations. Thus, in general, the situations featuring in different stories elicited different rank orderings of people whom the children perceived as most suitable sources of support. However, a few stories achieved moderately high correlations with other stories, most strikingly confiding a problem and support of self-esteem after a bad day. This indicates a similar choice of people nominated for their perceived value in providing help or support in these two situations. Other examples are comfort when scared and confiding a problem, and support in an embarrassing situation and confiding a problem, so those three scenarios tend to associate into a cluster. Intuitively, similarities in rankings between scenarios makes sense since children were selecting from their own limited choice of 10 special relationships who could be expected to give a range of support functions in differing situations.

However, some of the stories revealed that children hold quite firm, consistent views about the appropriateness of relationship functions to be obtained from different people. The most noticeable of these were the two stories concerning conflict which each had very low correlations with other stories, despite having high test-retest correlations. The story on comfort when ill had moderately high correlations with support for self-esteem after a bad day, but other correlations involving comfort when ill were generally low. Confiding a special secret also tended to have low correlations with other support scenarios. These low correlations suggest that the choice of people for providing supportive functions was different from one story to another.

These findings indicate that children can and do make considered selections of the appropriateness of people within their social network to fulfil relationship functions in accordance with a need presented by situations and that, child by child, their selection of appropriate people is consistent.

Further analysis was conducted to examine whether the children were similar to one another in their selection of particular types of relationships for particular needs. Although the children had slightly differing selections of relationships in their top 10 lists, many of the nominated relationships were similar. For example, all children nominated mother and father, siblings, best friends and at least one other family member such as an aunt or a grandparent. The 17 relationship types listed in Table 2 were sufficiently common across the sample to permit analysis. The category 'small pets' comprised guinea pigs, rabbits, hamsters and budgerigars.

To measure the degree of consistency in the way that the children ranked these relationships in response to the stories, intraclass correlation coefficients were computed (ICC(2,k) in the terminology of Shrout & Fleiss 1979). With one exception, these coefficients were substantial (range 0.56-0.90) and statistically significant (P < 0.01) for both session 1 and session 2 data. The exception was for story 3: self-esteem in an embarrassing situation which, although significant for session 1, fell just below statistical significance for session 2 (ICC = 0.4). Overall, there was a high degree of consistency across children in which particular kinds of relationships they selected for different stories. It would appear that children share similar perceptions both of the needs portrayed by the different stories and the appropriateness of particular relationships to elicit particular functions.

Table 2 shows which relationships were among the five most likely relationships to be selected for each story (based on mean ranks). Viewed by column this shows which relationships were most frequently selected for each story; viewed by row it shows how many stories each relationship was favoured for.

Younger siblings are ranked highly only in one or both of the stories concerning conflict, and seem not to be regarded as appropriate people from which to elicit help with a problem, comfort when ill or in a frightening situation, or to bolster esteem. This is much as would be expected. Older siblings are also highly ranked for both of the conflict situations, being seen as both annoying to the subjects and likely to perceive the subjects as a source of annoyance. Older brothers are also seen as appropriate sources of help in situations where a child may feel scared, perhaps as a function of them being an older person to accompany the child. Similarly, perhaps because of the nature of the problem described in the story in which a child was being bullied, older brothers were nominated as being of potential help. Older sisters were not regarded as appropriate for either of these stories.

Best friends were most likely to be the person with whom subjects would choose to share a special secret. Again, this is much as might be expected, as is the presence of best friends as high ranking in all but the conflict stories. Teachers figured only as potential helpers with a problem such as bullying and as sources of annoyance to subjects. Mothers were consistently ranked highly for all stories except that of being annoying to subjects. Fathers were less likely to be selected as appropriate for providing comfort when ill, or sharing a special secret, but as more appropriate for providing help when the child was scared, needed a boost to self-esteem or had a problem.

Some relationships nominated as being part of the top 10 list were not among the relationships most likely to be selected for any of the stories. These tended to be relationships that were not immediate family, such as uncles, aunts, and grandparents. In contrast, some relationships featured amongst many stories. This was especially so in the case of mothers, fathers and best friends and indicates a closeness of relationship that fulfils many needs across many situations. This reflects what Reid et al. (1989) referred to as generalist providers of support. Relationships fulfilling a narrower range of supports, or regarded as appropriate for only certain situations, are termed 'specialist' providers. Examples of specialist providers in this sample are teachers, as helpers with a bullying problem, and friends (not best friends) with whom a subject may share a secret.

Rankings given to pets suggest that children can realistically discriminate between functions that pets can or cannot provide. For example, no pets were nominated as being able to help with a bullying problem. Dogs were seen as a useful protector in a scary situation whereas cats were not. Perhaps because dogs are more active play companions than cats, dogs were also ranked highly for sharing a special secret. Small pets achieved high rankings only for self-esteem in an embarrassing situation-as non-judgemental relationships in front of whom a child could rehearse or practice and not feel silly if they made mistakes

However, the functions perceived as available from cats and dogs strongly suggests that children regard these as potential sources of comfort and/or esteem. Cats featured particularly highly as providers of comfort if a child was ill in bed. Dogs were also regarded as comforters when ill (although not as highly as cats) but were especially perceived as serving a supportive function in a frightening situation, and as playmates with whom to share a special secret. This would appear to suggest that pets play a specialist role in children's support networks that, although not confined to the provision of just one function, is limited by the child's appreciation of the animals' limitations. Provision of comfort, esteem and play appear to be the dominant roles for these pets.

It is also striking that cats and dogs consistently achieved higher rankings than many of the participants' human relationships, notably non-immediate family members such as aunts, uncles and grandparents. This would indicate that pets are regarded as close family members not only through residing in the same house, but through the functions they provide.

The aims of this study were to examine the significance of pets in children's social networks and the functions they may serve as perceived by the children, using a methodology designed to place pets in a context of the roles and functions available from human relationships in the children's social networks.

The methodology was found to be highly satisfactory in that it enabled children to generate their own choices of relationships, whereas other studies into relationship function that have often used relationship types prescribed by the researchers. The results leave little doubt that children are able to both identify special relationships and to discriminate between them to obtain appropriate support if needed. The methodology is therefore well suited to children who are too young for the use of self-report measures. The stories and illustrated story-boards were much enjoyed by the children (especially the stories concerning putting annoying people in the bin). Asking the children to make a sequence of 'best vs. the rest' judgements seemed more appropriate than asking them for a straight ranking of the relationships.

The results confirm the findings of Reid et al. (1989) that children's social networks contain both generalist and specialist support providers. In this study, generalists were identified as parents and best friends, and specialists as teachers and friends. Siblings were also identifiable as specialist providers of mutual annoyance and conflict although, with special regard to older siblings (especially brothers), there were possible functions as protectors or helpers with a problems.

The results also confirm the claims of Bryant (1985) that pets may assume significant relationships in children's social networks, although this was only apparent for cats and dogs, and add to our understanding of children's representations of these relationships. Cats were especially seen as comforting when a child was ill, whereas dogs were regarded as comforting during illness and supportive if the child were scared. Dogs were seen as companions/playmates for special ventures. Thus it would appear that dogs and cats offer special relationships for provision of psychological forms of support (i.e. making one feel better about oneself) but not for the more practical problems a child might have to deal with. The fact that cats and dogs frequently ranked higher than many human relationships suggests the value that children place on their pets and the functions they serve, and that studies examining children's networks would be advised to include any pets that may be owned.

This research was carried out while the authors were in receipt of financial support from the Waltham Centre for Pet Nutrition. June McNicholas is a Waltham Research Fellow. The authors thank Emma Myring for collecting the data.

Accepted for publication 13 September 2000

Legend for Chart:

B - Comfort when ill
C - Comfort when scared
D - Self-esteem (embarrassed)
E - Self-esteem (bad day)
F - Confiding a secret
G - Confiding a problem
H - Conflict: (annoys you)
I - Conflict: (you annoy)

            A                 B       C       D       E
                              F       G       H       I

Comfort when ill             0.49    0.22    0.25    0.41
                             0.24    0.29   -0.09    0.02

Comfort when scared          0.12    0.48    0.33    0.32
                             0.34    0.31   -0.14    0.18

Self-esteem (embarrassed)    0.10    0.19    0.43    0.27
                             0.27    0.32   -0.06    0.14

Self-esteem (bad day)        0.34    0.30    0.45    0.57
                             0.25    0.51   -0.21    0.18

Confiding a secret           0.27    0.29    0.27    0.21
                             0.54    0.18   -0.04    0.05

Confiding a problem          0.16    0.46    0.45    0.57
                             0.24    0.63   -0.14    0.25

Conflict: (annoys you)      -0.14   -0.05   -0.19   -0.24
                            -0.11   -0.12    0.66    0.25

Conflict: (you annoy)        0.15    0.21    0.11    0.24
                             0.08    0.25    0.30    0.81

The correlations on the diagonal of the table are
test-retest correlations for the same story across the two
sessions. The correlations in the off-diagonal cells are between
different stories, either in session 1 (in the upper-right part
of the table) or in session 2 (lower-left). The correlation
coefficients were computed as pooled within-subjects correlations
with 104 degrees of freedom (based on 22 subjects and five
observations per subject). We suggest r = 0.3 as conservative
criterion for statistical significance.

Legend for Chart:

A - Relationship
B - Story Comfort when ill
C - Sotry Comfort when scared
D - Story Self-esteem (embarrassed)
E - Story Self-esteem (bad day)
F - Story Confiding a secret
G - Story Confiding a problem
H - Story Conflict: (annoys you)
I - Story Conflict: (you annoy)

     A        B   C   D   E   F   G   H   I

Mum           *   *   *   *   *   *       *

Dad               *   *   *       *       *

Best friend   *   *   *   *       *

Brother[O]        *       *       *   *   *

Dog           *   *           *

Sister[Y]                             *   *

Cat           *

Cousin        *               *

Brother[Y]                            *

Sister[O]                             *   *

Friend                        *

Teacher               *           *   *

Aunt

Grandma                   *

Grandad

Small pet             *

Uncle

[O] = older sibling

[Y] = younger sibling.

References

Bryant, B. K. (1985) The Neighbourhood Walk: Sources of Support in Middle Childhood. Monographs of the Society for Research in Child Development, 50, 1-116 (Serial no. 210). Society for Research in Child Development, London.

Bryant, B. K. (1990) The richness of the child-pet relationship: a consideration of both benefits and costs of pets to children. Anthrozoos, 3, 253-261.

Covert, A. M., Whiten, A. P., Keith, J. & Nelson, C. (1985) Pets, early adolescents and families. Marriage and Family Review, 8, 95-108.

Davis, J. H. & Juhasz, A. M. (1995) The preadolescent pet friendship bond. Anthrozoos, 8, 78-82.

Dunn, J. (1993) Young Children's Close Relationships: Beyond Attachment. Sage, Newbury Park, CA.

Endenburg, N. (1995) The attachment of people to companion animals. Anthrozoos, 21, 83-89.

Furman, W. & Buhrmester, D. (1985) Children's perceptions of the personal relationships in their social networks. Developmental Psychology, 21, 1016-1024.

Melson, G. F. (1998) The role of companion animals in human development. In: Companion Animals in Human Health (eds C. C. Wilson & D. C. Turner), Sage, Thousand Oaks, CA.

Rehm, N. (1993) Dog keeping in the family. In: Proceedings of the International Congress on Applied Ethology, Berlin 1993 (eds M. Nichelmann, H. K. Wierenga & S. Braun), Humboldt University, Berlin.

Reid, M., Landesman, S., Treder, R. & Jaccard, J. (1989) 'My family and friends': six-to twelve-year-old children's perceptions of social support. Child Development, 60, 896-910.

Shrout, P. E. & Fleiss, J. L. (1979) Intraclass correlations: uses in assessing rater reliability. Psychological Bulletin, 86, 420-428.

Van Houte, B. A. & Jarvis, P. A. (1995) The role of pets in preadolescent psychosocial development. Journal of Applied Developmental Psychology, 16, 463-479.

~~~~~~~~

By J McNicholas, Department of Psychology University of Warwick Coventry CV4 7AL UK and G M Collis, Department of Psychology, University of Warwick, Coventry, UK