IEEE Robotics & Automation Magazine - March 2011 - 35

It may be that a robot that seems to resemble us, or to
respond to us, will inevitably be anthropomorphized to
some degree. Designing robots to encourage anthropomorphic attributions could therefore be viewed as an unethical
form of deception. However, in that case, giving any object a
human or animallike appearance could also be seen as
deception. It seems too extreme to suggest that dolls, puppets, and statues should no longer be made or played with.
People, in general, and children, in particular, exhibit anthropomorphic behavior much of the time. Anthropomorphic design occurs in many more areas than robotics, from
Alessi bottle openers to car grilles and even pet rocks [17].
Rather than objecting to all such uses, it makes more sense
to focus our ethical concern on those situations in which
anthropomorphic design seems likely to lead to negative
consequences for human welfare. Some such consequences
are considered in the following section.
A further cause for concern is that there are reasons to
expect the vulnerable youngest and eldest members of
society to be more likely to be affected by anthropomorphism. Both have a strong need for social contact, and both
may lack knowledge of the technology underlying the
apparent responsiveness of interactive robots. Both these
factors have been argued to increase the tendency to be
anthropomorphic in recent accounts [16].
Epley et al. [16] argue that the tendency to anthropomorphize nonhuman agents depends on three psychological determinants: the accessibility and applicability of
anthropocentric knowledge, the motivation to explain and
understand the behavior of other agents, and the desire for
social contact. Their argument is backed up by extensive
experimental evidence, of which a few examples are cited
here. Various factors can be shown to affect the accessibility and applicability of anthropocentric knowledge: for
example, greater similarity between the appearance and
behavior of an entity and humans, or animals, can increase
the degree of anthropomorphism and empathy shown
toward it. Thus, DiSalvo et al. [18] found that robots are
anthropomorphized more readily when given humanlike
faces and bodies. The idea that anthropomorphism is
stronger when there is a need to explain is supported by the
evidence that shows that unpredictable behavior increases
the tendency for anthropomorphic explanations [19]. Finally,
in accord with the desire for social contact determinant,
experimental manipulations show that, when feelings of
loneliness are induced, people are more likely to anthropomorphize pets and gadgets [20].
This account of anthropomorphism can be used to
argue that both the very young and the very old may be
more likely than other age groups to be anthropomorphic
and less able to understand the limited ability of robots to
understand and empathize. Both groups have a strong
desire for social contact: babies (because they are innately
predisposed to look for human social contact) and the
elderly (because they are often lonely). In addition, both
are likely to lack knowledge about how robots work.

Infants and young children are not clear about the differences between living and nonliving entities [21]. Elderly
people with Alzheimer's may not be able to understand the
mechanisms underlying robot behavior. Both groups might
be more prepared to form relationships with robots and
robot pets designed to give the illusion of sentience than
other groups of the population.
Both the young and old may show a stronger tendency to
anthropomorphize robot companions and pets, but whether
or not this is amounts to an ethical problem depends in part
on what the consequences of such anthropomorphism might
be. We consider these in the following sections.
Likely Consequences
for Robots and the Elderly
One negative consequence of an elderly person imagining
that they have a relationship with a robot might be an
increase in their level of anxiety-they might think that
they had to look after the robot, even at the expense of their
own well-being. Observers and relatives of a confused old
person looking after a robot pet might see it as depriving
their relative of dignity and infantilizing them.
Similar points have been made in the context of the doll
therapy that has been undertaken with those with Alzheimer's disease. Positive effects have been found from doll
therapy, where dolls are given to clients to stimulate memories of a rewarding life role, especially that of a parent,
and to act as a focus for reminiscence and conversation
[15]. However, ethical objections have been raised to the
effect that doll therapy infantilizes the elderly [15].
Studies have shown that clients with dementia engaged
in doll therapy tend to believe that their dolls are real babies.
When Mackenzie et al. [22] questioned the care workers in
homes where doll therapy had been tried, they discovered
that some residents would put the doll's interests before
their own as one would with a real baby. They also found
that some caregivers, visiting relatives, and fellow residents
saw the doll therapy as demeaning and patronizing.
Looking after robot pets could be seen to similarly
infantilize elderly people, although a mitigating factor is
that robots can be seen as cool gadgets in a way that dolls
are not. Another possible negative consequence is that the
presence of a robot might result in a reduction in the level
of social interaction an elderly person experiences. An outcome in which an elderly person chose to spend time with
the robot rather than taking part in social interactions with
humans would be unwelcome. Similarly, if other people
were to assume that the social needs of an elderly person
were being taken care of by the robot and so interacted less
with them, that would also be a problem.
On the other hand, there are reasons to expect some
positive outcomes. Various studies have found evidence
that the elderly can benefit from interacting with robot
companions. The positive effects are said to be similar to
those obtained from animal-assisted therapy [23]. For
instance, Kanamori et al. [24] showed various improvements
MARCH 2011

IEEE ROBOTICS & AUTOMATION MAGAZINE

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2011