A peer pressure experiment: Recreation of the Asch conformity
experiment with robots.
J
¨
urgen Brandstetter
1
, P
´
eter R
´
acz
2
, Clay Beckner
2
, Eduardo B. Sandoval
1
, Jennifer Hay
2
, Christoph Bartneck
1
Abstract— The question put forward in this paper is whether
robots can create conformity by means of group pressure. We
recreate and expand on a classic social psychology experiment
by Solomon Asch, so as to explore three main dimensions. First,
we wanted to know whether robots can prompt conformity in
human subjects, and whether there is a significant difference
between the degree to which individuals conform to a group
of robots as opposed to a group of humans. Secondly we ask
whether group pressure (from human or robot peers) can exert
influence in verbal judgments, analogously to the influence on
visual judgments that is known from previous research [3], [2].
Thirdly, we investigate whether the level of conformity differs
between an ambiguous situation and a non-ambiguous situation.
Our results show that in both visual and verbal tasks,
participants exhibit conformity with human peers, but not
with robot peers. The social influence of robot peers is not a
significant predictor of verbal or visual judgments in our tasks.
Furthermore, the level of conformity is significantly higher in
an ambiguous (unclear) situation.
I. INTRODUCTION
When we look at the current development of human-
like social robots, it is possible to predict a future where
robots help out in the office, teach children in the class-
room, become companions, work in advertising, or help in
our households [5], [14], [20]. During the last few years,
there has been an observable trend towards service robots
outflanking industrial robots in production volume. Whereas
all sold industrial robots total 2.3 million units, more than
2.5 million units of service robots were sold in 2011 alone,
with an estimated 16 million further units to be sold between
2012 and 2015 [24]. The importance of robots is not only
recognized by the science community and the industry,
but also by governments. For example, the New Zealand
government has put the development of robots on their main
future agenda.
With this increased presence of robots in domestic life, the
question remains as to how these robots will be regarded by
humans—as mindless machines, as subservients, or as peers?
Previous work demonstrates that humans treat computers
socially, for instance, by tempering unpleasant feedback to a
computer, apparently so as to be more polite [19], [9]. Such
findings raise the question of whether it is possible for robots
to prompt human conformity.
1
Human Interface Technology Lab, University of Canterbury, P.O. Box
4800, Christchurch, 8140, New Zealand
2
New Zealand Institute of Language, Brain and Behaviour, University of
Canterbury, P.O. Box 4800, Christchurch, 8140, New Zealand
Please direct correspondence to J
¨
urgen Brandstetter at
P
´
eter R
´
To investigate the persuasive power of robots, we ex-
pand on two landmark psychology experiments. The first
experiment was conducted by Muzafer Sherif, who studied a
conformity effect in an ambiguous situation. The experiment
builds on the autokinetic effect, a phenomenon in which a
person perceives a sudden movement of a light point when
no reference is given. Sherif created this effect by sitting
participants in a dark room with one small light point visible.
The participants had to look at this point, and had to say
how much the light point moved. However, the light never
actually moved, but the autokinetic effect created an illusory
sudden movement. This movement is perceived differently
by different people. After he did the experiment with one
person, Sherif set two or three people in the same room
and asked them to say out loud how much the point moved.
Astonishingly, after three rounds all the participants said the
same number, even though everyone perceived a different
movement. This effect is called informational conformity or
social proof and describes the effect in which people in an
uncertain situation look at their neighbors to see what is
probably the right answer and conform with them. [7], [21],
[15].
A second experiment is reported by Solomon Asch in his
influential paper “Effects of group pressure upon the mod-
ification and distortion of judgements”. Asch builds on the
findings of Sherif to find out whether humans also conform in
non-ambiguous situations. His experiment simulated a simple
visual line test. The participant saw three lines of different
heights, labelled A, B, C on the left side of a board, and one
reference line on the right side labelled with “?” (see Figure
1). The task was to say which line matched the reference line.
When the participants were alone in a room to perform the
test they almost always gave the correct answer. In a second
round Asch placed the participants in the same room with
other participants who were all actors and who all gave the
wrong answer. Even though the real participant – presumably
– knew the correct answer, in 32% of all tasks the participants
went along with the group if the group size was bigger than
four. In this case, Asch provided evidence that conformity
is not only influential in ambiguous situations but also in
non-ambiguous ones [2], [3], [4].
In the current study, we recreate the Asch experiment and
use aspects of the Sherif experiment as a model for further
research. We investigate a visual judgment task along the
lines of Asch [2], [3], [4], and extend the methodology
to a verbal production task. Verbal tasks may provide an
especially fruitful domain for conformity studies, because
language is inherently social, and speakers influence one
2014 IEEE/RSJ International Conference on
Intelligent Robots and Systems (IROS 2014)
September 14-18, 2014, Chicago, IL, USA
978-1-4799-6934-0/14/$31.00 ©2014 IEEE 1335