CARING FOR SHARING
Caring for Sharing: How Attachment Styles Modulate Communal Cues of Physical Warmth
VU University, Amsterdam
Radboud University, Nijmegen
University of Copenhagen & Harvard University
Thomas W. Schubert
Manuscript in press
Social Psychology: Special Issue on the Fundamental Dimensions of Social Perception
Hans IJzerman, Department of Social Psychology, School of Social & Behavioral Sciences,
Tilburg University, The Netherlands; Johan Karremans, Department of Social & Cultural
Psychology, Radboud University, Nijmegen; Lotte Thomsen, Department of Psychology,
University of Copenhagen & Department of Psychology, Harvard University; Thomas
Schubert, CIS/ISCTE, Lisbon. Correspondence concerning this article should be addressed to
firstname.lastname@example.org. We would like to thank Emma Landstra for her hard work in collecting
data and the Tilburg Hardcore Labgroup, the editor Bogdan Wojciszke and two anonymous
reviewers for their helpful comments. Addresses of the schools where our data was collected
are available upon request from the first author.
CARING FOR SHARING
Does physical warmth lead to caring and sharing? Research suggests that it does; physically
warm versus cold conditions induce pro-social behaviors and cognitions. Importantly, earlier
research has not traced the developmental origins of the association between physical warmth
and affection. The association between physical warmth and sharing may be captured in
specific cognitive models of close social relations, often referred to as attachment styles. In
line with this notion and using a dictator game set-up, the current study demonstrates that
children who relate to their friends in the manner of a secure attachment style are more
generous toward their peers in warm as compared to cold conditions. This effect was absent
for children who relate to friends in the manner of an insecure attachment style, but, notably,
these children not just always shared less: They allocated more stickers to a friend than to a
stranger. These findings provide an important first step to understand how fundamental
embodied relations develop early in life. We discuss broader implications for grounded
cognition and person perception.
Keywords: attachment, grounded cognition, relational models, physical warmth,
CARING FOR SHARING
Caring for sharing: How attachment styles modulate communal cues of physical warmth
Children’s experiences with their caretakers differ substantially. Previous research has
identified responsiveness of caretakers as an important variable that determines behavior in
social relations by contributing to an attachment style developed by the child. The concepts of
caretaker responsiveness and attachment styles may suggest a rather abstract and high-level
concept. However, what co-varies with responsiveness may often be very concrete behavior
such as touching, nursing, and feeding that all co-occur with providing bodily warmth. Such
concrete dimensions of social interactions have recently received increased empirical
attention in the literature on adult and infant social cognition (e.g., Häfner & IJzerman, 2011).
We argue here that attachment styles also have an embodied dimension that tunes social
behavior to embodied cues such as warmth.
Subtle manipulations of physical warmth have been linked to pro-social behaviors and
cognitions as compared to physically cold conditions. For instance, adult participants in
physically warm as compared to cold conditions rated a third party as more sociable and were
themselves more generous (Williams & Bargh, 2008; see also Asch, 1946, 1958). Similarly,
physically warmer conditions attune adults to focus on communal, affiliate relationships (like
that between partners or between caregiver and infant). That is, adults in physically warm
conditions use more language associated with relationships, perceive more relationships in
their environment, and construe their selves as having greater integration with others
(IJzerman & Semin, 2009). Vice versa, the lack of social relationships leads people to
perceive their environment literally as physically colder (IJzerman & Semin, 2010; Zhong &
The theoretical accounts underlying these experiments typically assume that the link
between warmth and pro-social cognition is the result of life-long over-learning facilitated by
an evolved proclivity (Williams & Bargh, 2008). Yet, little is known about the developmental
CARING FOR SHARING
path of this link, and still less about its generality across individuals. In the current work, we
trace the causal role of an early cognitive link in children, and find that children who did not
develop a secure internal model of relationships towards friends also seem not to show this
General Relations between Attachment, Touch, and Warmth
Research linking warmth and communal behavior is based on the idea that from
infancy on, physical warmth and experienced affection co-vary. This is argued to lead to a
learned association of warmth and communion, possibly facilitated by an evolutionary
adaptation (Williams, Huang, & Bargh, 2009). These arguments concur with Harlow’s (1958)
work on touch. He suggested that close, physical contact throughout infancy is necessary for
healthy psychological functioning: Young rhesus monkeys preferred a soft surrogate mother
made of terrycloth over a surrogate mother made of wire, and those monkeys raised with a
wire mother (as compared to the terrycloth mother) experienced greater stress, were less
playful, and were less likely to explore the environment in new situations. This may suggest
that close bodily contact with a care-giver, which is often associated with physical warmth,
may indeed be an evolved need (cf. MacDonald, 1992).
Experiences of infants with their caregivers, however, can vary considerably. One
prominent account on how people learn to interact with close and communal others,
attachment theory (Bowlby, 1969), suggests that infants come to learn if their caregiver is
likely to be responsive and available to meet their needs. Early interactions with communal
others (like the caregiver) inform infants on specific manners in which they are expected to
behave towards self and others. Such early interactions are assumed to be the basis for more
generalized cognitive models of relationships and are commonly referred to as attachment
styles (cf. Ainsworth, Blehar, Waters, & Wall, 1978). Those infants who are securely attached
expect their caregivers to be available, and when upset, they are more easily comforted. In
CARING FOR SHARING
contrast, insecurely attached infants do not trust their caregiver will be reliably available for
their comfort and protection and are not easily comforted (Johnson, Dweck, & Chen, 2007).
The development of attachment styles has been linked to embodied cues experienced
in infancy early on, especially bodily contact (Bowlby, 1969). Ainsworth (1979) reported that
mothers of avoidant babies showed more frequently signs of an aversion to close bodily
contact. Such avoidance of bodily contact frustrates the child’s attempts to establish secure
bodily contact when being motivated to do so, and changes their own approach behavior
(Anisfeld, Casper, Nozyce, & Cunningham, 1990; Blehar, Lieberman, & Ainsworth, 1977).
This suggests that early embodied cues (in contrast to higher order cognitive cues, like the
differentiation between friends and strangers) - such as the inherently related touch and
physical warmth – may form critical parts of the meaning of cognitive models of attachment.
In other words, we argue that an evolutionary adaptation to associate warmth and
caring is likely to afford the child with a crucial learning opportunity; however, caregiver
behavior such as that leading to an avoidant attachment style may be able to disrupt the
acquisition of this link. If so, effects of physical warmth on behavior should be modulated by
The Present Research: Attachment, Friendship, and Warmth
In the present work, we focus on one specific communal relation, namely friendship
relations. Friendships feature warmth and touch less prominently than the first, crucial parent-
infant bond; nevertheless, anthropological and psychological work alike suggests that
friendship relationships are often considered as communal or even very similar to kinship
relations (Ackerman, Kenrick, & Schaller, 2007; Adams & Plaut, 2003). Indeed, friends are
often considered as “intimate associates”; the relations between friends are characterized by
commitment, cooperation, affection, and prosocial support (Hartup, 1980). Among adults,
secure attachment in general provides a foundation for compassion and care-giving
CARING FOR SHARING
(Mikulincer & Shaver, 2005). Being insecurely attached, in contrast, hampers compassionate
caregiving (e.g., Carnelley, Pietromonaco, & Jaffe, 1996; Collins & Feeney, 2000).
Self-reported attachment styles to friends should thus moderate how young children
respond to physical warmth in their behavior toward their peers: Children with different
orientations to close and communal friend relations, as measured through an adapted measure
of their attachment styles to friends, should differ in the way they respond to physical warmth
with caring actions, because their experiences with the implied communal friendships differ.
To test this interaction hypothesis, we assessed children’s self-reported attachment
styles towards friends, manipulated room temperature, and measured how many stickers the
children shared in a dictator-game (e.g., Forsythe, Horowitz, Savin, Sefton, 1994) with
another, not present, child under these conditions. Note that we thereby manipulated physical
warmth independently from touch, around comfortable ranges. Furthermore, communal
behaviors with respect to friendship relations do not afford close physical warmth to the same
degree as parent-child relations. Together, this assures a strong test of our hypothesis.
Whereas securely attached children should respond intuitively to physical warmth by acting
communally and altruistically to meet other’s needs, this should not be the case for insecurely
attached children who do not share the experience that the needs of self and others are met in
(physically) warm, communal relationships. This interaction hypothesis should be absent for
known, learned relationships (such as the difference between friends and strangers).
Sixty (43.3% female, Mage = 5.08, SDage = .59, all native Dutch) children were
recruited from Kindergarten classes at two elementary schools in Abcoude, the Netherlands.
For all children, we obtained informed consent from their parents/caregivers to participate.
All children received ten stickers and twenty balloons at the end of the day of which they
CARING FOR SHARING
participated. One child was excluded from the final analysis as the task was too difficult for
him to complete. All children were randomly allocated to the physically cold (N = 29) or
warm (N = 30) condition.
Procedure and Design
The children were accompanied from their classroom to a room, located in their
elementary school, which was either cooled down (cold condition; 15-19 °C) or heated up
(warm condition; 21-26 °C) beforehand (cf. IJzerman & Semin, 2009). At the beginning of
the session, the experimenter, who was blind to the hypotheses, asked them 15 questions from
our attachment scale in regards to friendship relations (derived from Bartholomew &
Horowitz, 1991). We simplified the original scale for usage with 4-6 year old children. All
children were trained in using simplified Likert-scales in advance with intuitive and easy
questions (e.g. “Do you like Brussels sprouts?”). The questions were administered orally and
the children could point to a 7-point scale, ranging from ‘not at all’ to ‘very much’ (see Figure
1 in Appendix A).
In some cases the children misunderstood; in those cases our experimenter re-
explained the questions and the way in which the children could answer. Our scale consisted
of four subscales, namely Dismissive Avoidant (4 items, e.g., “Do you feel at ease without
having good friends”), Fearful Avoidant (4 items, e.g., “Do you find it difficult to be in need
of other children’s help?”), Securely Attached (5 items; “Do you find it easy to become good
friends with other children?”), and Anxious Preoccupied (3 items; e.g., “How do you feel
without good friends? “). One item was used both for Securely Attached and Fearful Avoidant
(but reverse-scored; see Appendix A).1
1Children of the age in our sample might show considerable variance in terms of how to use the scale. We
controlled for this in two ways. First, our experimenter explained questions the children might not intuitively
understand. Second, we refrained from using the attachment scale to predict responses on physical warmth.
Instead, we categorized children on the basis of their highest average score on our attachment scale, reducing the
relevance of children’s idiosyncratic use of the scale.
CARING FOR SHARING
Subsequently, the children were shown ten equally sized stickers (or twenty balloons)
and asked how many they wanted to keep for themselves and how many they wanted to share
with an ostensible other child next door. In reality (but unbeknownst to the child), this was a
fictitious child, of the same age and sex as the participant, not acquainted and would not be
met in the future. All children believed that the other child truly existed. Our experimenter
told the children that the other child would not get the option to receive stickers (or balloons).
She then asked the same question with ten new stickers, but children were now asked if they
wanted to keep all ten for themselves or if they preferred to share some with their best friend
instead (of which we recorded the name in advance). The order stranger-friend was
counterbalanced across children. At the end, children were asked not to tell their classmates
about the content of the experiment.
We classified children as securely or insecurely attached on the basis of their highest
averaged score on one of the subscale of our attachment scale (62.7% securely attached).
Insecurely attached children were those who scored highest on a subscale other than securely
attached (37.3% insecurely attached, this was composed of 8.5% Dismissive Avoidant; 6.8%
Fearful Avoidant; 20.3% Anxious Preoccupied; 1.7% scored highest on all three insecure
styles). In the case of an equal score in the secure subscale as compared to one of the insecure
subscales, we averaged the three insecure subscales and re-compared. This happened in four
cases. Subsequently, we analyzed our results according to a 2(between: temperature: cold vs.
warm) x 2(between: attachment style: secure vs. insecure) x 2(within: friend vs. stranger)
mixed-models design. There was no reliable effect of order on giving to friend or stranger
(p=.551). In addition, we analyzed whether temperature affected the children’s answers to the
attachment questionnaire. A logistic regression of temperature on attachment styles showed no effect
CARING FOR SHARING
As we expected, known relationships like differences between friends and strangers
showed a simple main effect: Children were more willing to share with their friends (M=2.97,
SD=2.33) than with strangers (M =.98, SD =1.74), F(3, 55)=34.38, p<.001, Cohen’s d = .97.
Notably, no interaction effect emerged between attachment style and friend-stranger factors,
F<1. Insecurely attached children thus also meaningfully adapted their sharing behavior to the
known nature of the relationship. Given that there were no further interaction effects
involving differences between friends and strangers (all ps>.250), we focused on the
hypothesized between-participants temperature X attachment style analysis by calculating an
average score for stickers given to friends and strangers.
In general, securely attached children gave more stickers than insecurely attached
children, F(1, 55)=5.34, p=.023, ηp2=.090, confirming the idea that securely attached children
are more care-‘giving’ than insecurely attached children. There was no main effect of
temperature, F<1. Importantly, the main effect of attachment was characterized by a
significant interaction of temperature and attachment style, F(3, 55)=5.69, p=. 020, ηp2=.094.
Securely attached children gave significantly more stickers in the warm (M=2.89, SD=1.70)
as compared to the cold (M=1.75, SD=1.53) condition, t(58)=3.35, p=.001, Cohen’s d = .71.
There were no significant differences for insecurely attached children, t(58)=-1.23, p=-.214
(warm, M=.95, SD=1.08; cold, M=1.77, SD=1.59). This confirms our hypothesis that
securely attached children should become more pro-social in warm as compared to cold
conditions, but that this effect should be absent for insecurely attached children.
We examined altruistic behavior of securely and insecurely attached children in either
physically cold or warm ambient conditions. In a dictator game, securely attached children
donated more stickers to two other children in physically warm rather than cold conditions.
This is remarkable from several perspectives. First, we provide evidence that the link between
CARING FOR SHARING
warmth and prosociality indeed goes back to early childhood, as assumed in earlier work
(IJzerman & Semin, 2009, 2010; Williams et al., 2009). It could have been possible that
adults show effects of warmth on prosociality because they learned the metaphor of kindness
and warmth, and that experiencing warmth influences kindness-related behavior because of
the representation of this metaphor (Landau, Meier, & Keefer, 2011). However, the fact that
we find the same effect with young children supports the notion that the effects found in
adults are due to the formation of modal knowledge structures during development (IJzerman
& Koole, 2011).
Furthermore, our findings add to the literature on embodied cognition, which so far
has mainly been concerned with main effects of embodied cues. We show that a higher order
concept such as attachment style developed during infancy can moderate the effect of
embodied cues (cf. Meier, Sellbom, & Wygant, 2008; Schubert & Koole, 2009). Going
beyond previous work showing main effects of warmth, we find that only securely attached
children’s giving behavior was positively affected by physical warmth, while insecurely
attached children did not show this pattern. Together, our findings indicate that insecurely
attached children do not respond to physical warmth with caring actions to the same degree as
securely attached children. In other words, one’s experiences with and expectations for a
given communal relationship determine how one intuitively responds to embodied or physical
communal cues. Moreover, because communal behaviors with respect to friendship relations
do not afford close physical warmth to the same degree as parent-child relations, the present
findings strongly suggest that the warmth-caring association underpin the meaning of
communal relations at large, irrespectively of learned associations in the specific social
We developed our argument based on the assumption that different caregiver
behaviors lead to different attachment styles and different associations of caring and
CARING FOR SHARING
embodied cues like physical warmth, which are of primary importance during early
experiences of being cared for (Ainsworth 1979). Infants face the formidable task of learning
who has what kind of relationship with one another in their social world. To learn this, they
must somehow know what basic kinds of social relationships to look for and how to recognize
them, that is, what are their physical or embodied cues. Basic components of communal
relations might help infants to construct meaningful internal working models of attachment on
the basis of their early life experiences. Strong learning accounts might underestimate the
bodily constraints and social coordination devices people are afforded with (IJzerman &
Cohen, 2011; IJzerman & Koole, 2011; Fiske, Thomsen, & Thein, 2009; Thomsen & Carey,
Recent evidence suggests that people may possess specific systems that link relative
size to dominance and physical closeness to communal relations among pre-verbal and just-
linguistic infants (Thomsen, Frankenhuis, Ingold-Smith & Carey, 2011; Over & Carpenter,
2009). We propose here that physical warmth also forms part of people’s core systems so as
to affect behavior in communal friendship relations. Importantly, the fact that children who
are insecurely attached to their friends do not respond to physical warmth with communal
actions towards friends indicates that the effect of embodied, physical cues is not absolute.
Instead, it hinges crucially on the relational meaning a given relation has for the child. This
depends on his/her prior experiences with and expectations to the relationship, as indexed by
attachment style. In other words, if the child does not experience friendships as secure and
successful communal relationships, she/he will not act communally towards friends based on
embodied, communal cues.
As far as we know, the present findings are the first to demonstrate that the
relationship between warmth and prosocial behavior is present in early childhood, and
moderated by attachment style. As such, the findings provide some initial insight into the
CARING FOR SHARING
origins of the warmth-prosociality link. However, we should acknowledge several limitations
and unaddressed issues. First, our findings do not show a simple main effect, as found in other
studies (IJzerman & Semin, 2009; Wiliams & Bargh, 2008). However, these other studies
focused on a selected group of college students, whereas our sample may have included a
broader range of children. In both samples used by the previous authors, the groups tend to be
a relatively selected group of college students. Recent research addresses how these groups of
educated students differ substantially from other samples across different tasks (Henrich,
Heine, & Norenzayan, 2009). Our research suggests that different samples may differ in
attachment styles, which may be the focus of future investigations.
Coming to the issue of differences across groups, one may argue that cross-cultural
climatic patterns do not converge with our findings (e.g., Van IJzendoorn & Kroonenberg,
1988). Yet, in our research we were interested in subtle temperature manipulations around
comfortable ranges, which seem qualitatively different from climatic differences in
temperature. The present research thus does not allow for interpretations of meanings of
findings from cross-cultural climatic patterns.
Furthermore, one may argue that mood might have affected our findings. However,
studies investigating subtle temperature manipulations consistently fail to find effects on
mood, and in fact argue against such a general “halo” effect (IJzerman & Semin, 2009;
Williams & Bargh, 2008). Finally, we were interested in a first exploration of attachment
style, thereby dividing children into secure and insecure groups. It is an important challenge
to discover how different insecure attachment styles may fare towards the experience of
physical warmth. Extending these findings to the other attachment styles will directly
contribute to the meaning of physical warmth in close relationships.
It is interesting to speculate how these findings play a role in more complex
cognitions. Extent research has addressed how people perceive others on an individual and
CARING FOR SHARING
group level, utilizing the metaphor of psychological warmth to describe how one feels about
others (Abele & Wojciszke, 2007; Fiske, Cuddy, & Glick, 2007; Wojciszke & Abele, 2008;
Wojciszke, Abele, & Baryla, 2009).Williams and Bargh (2008) confirmed Asch’ (1958) early
suspicions that the metaphor of psychological warmth is firmly grounded in the physical
experience of warmth, by showing how experiencing physical warmth increases how sociable
people perceive others to be. Other, recent work, shows that this link is bi-directional, in that
reminding people of own and other’s positive communal traits increases perceptions of
ambient temperature, independent of self-esteem and higher order cognitions of relationships
(Szymkow, Chandler, IJzerman, Parchukowski, & Wojciszke, 2012).
But how do people learn to think of others as warm or cold? Also here, one may take a
strong learning approach (Lakoff & Johnson, 1999; Landau, Meier, & Keefer, 2010), but as
we mentioned above, strong learning accounts overlook the evolved proclivities people
become equipped with in order to engage in relationships (Caporael, 1997; Fiske, 1991, 1992;
IJzerman, Gallucci, Pouw, Weiβgerber, Van Doesum, & Williams, 2012; Thomsen & Carey,
2012). Examining the combinatorial logic of learning accounts scaffolded onto evolved
proclivities through physical warmth is vital in understanding why psychological warmth has
become such an important dimension in perceiving others, and for analyzing how the
dimension differs across individuals and groups.
To conclude, co-experiencing physical warmth with communal bonds reinforces
socially meaningful experiences. The interaction of the body with the social world provides
important information for how people learn to engage effectively and adaptively in
interpersonal relations. In the current experiment we manipulated room temperature, but in
actual situations a temperature cue would be typically instantiated by a warm touch – to
which, as we see, securely attached children could adequately respond. Our results therefore
CARING FOR SHARING
form an important step as to what bodily constraints and social coordination devices are
essential in going “beyond the information given” when constructing social relations.
CARING FOR SHARING Download full-text
Abele, A. E., & Wojciszke, B. (2007). Agency and communion from the perspective of self
versus others. Journal of Personality and Social Psychology, 93, 751 – 763.
Ainsworth, M. D. (1979). Infant--mother attachment. The American psychologist, 34, 932-
Ainsworth, M. D., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment.
Hillsdale, NJ: Erlbaum.
Anisfeld, E., Casper, V., Nozyce, M., & Cunningham, N. (1990). Does infant carrying
promote attachment? An experimental study of the effects of increased physical
contact on the development of attachment. Child Development, 61, 1617 – 1627.
Asch, S. E. (1946). Forming impressions of personality. The Journal of Abnormal and Social
Psychology, 41, 258-290.
Asch, S. E. (1958). The metaphor: A psychological inquiry. In R. Tagiuri & L. Petrullo
(Eds.), Person perception and interpersonal behavior (pp. 86-94). Stanford, CA:
Stanford University Press.
Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test
of a four-category model. Journal of Personality and Social Psychology, 61, 226-244.
Blehar, M. C., Lieberman, A. F., & Ainsworth, M. D. S. (1977). Early face-to-face interaction
and its relation to later infant-mother attachment. Child Development, 48, 182-194.
Bowlby, J. (1969). Attachment and Loss. London: Hogarth Press.
Carey, S. (2009). The Origins of Concepts. Oxford University Press.
Carnelley, K. B., Pietromonaco, P. R., & Jaffe, K. (1996). Attachment, caregiving, and
relationship functioning: Effects of self and partner. Personal Relationships, 3, 257 –