Review about “Large Stakes and Big Mistakes” paper
Is it always an increase in motivation that lead to an improvement in performance? Usually, in the
common thinking the answer for this question is: yes, but it is not really so. We will see in some cases
that higher monetary incentives lead to worse performance. Intuitively, we could think that (1) an
increase of performance-contingent incentives produces, as a result, an increase of motivation and
effort and (2) this improvement of motivation/effort lead to a better performance. Since this second
assumption is so deeply held, the research has examined more the connection between pay and
motivation/effort than motivation/effort and performance. Now, thanks to some evidence, we have
seen that, in some situations, an increase of motivation and effort beyond a threshold level can lead to
a decrement of performance. This phenomenon is known as “Choking under Pressure”. Although there
is a positive correlation between performance and effort, in some cases we could also experience the
opposite relationship due to psychological mechanisms like increased arousal, shifting mental
processes from “automatic” to “controlled” etc. About the increased arousal, the “Yankes-Dodson law”
(Yankes and Dodson, 1908) shows that there is an optimal level of arousal for doing tasks, beyond
which, there is a decrement in productivity (demonstrated by the rat experiment). Since arousal is
strictly correlated with motivation and performance, the “Yankes-Dodson law” suggests that an
increase of motivation exceeding the optimal level, in some situations, can lead to a supra-optimal
level of arousal which bring a decline of performance.
Another negative effect of an increase of incentive could be, involuntarily, shifting from an “automatic”
behaviour to a “controlled” one. This leads also to a decrease in performance (see Dandy, Brewer and
Tottman, 2001, for an example about sport).
There could also be another mechanism that could transform an increase in motivation to decrease in
performance and this involves the focus of attention. McGraw and McCullers (1979) show that:
introducing a monetary reward for tasks that involved problem solving, for example, could induce
detrimental effects on performance, a large incentive can simply occupy the mind of the person
distracting her from the task. For economics, it is interesting to study how the level of performance-
contingent monetary incentives affects the performance pressure. The experiment in this paper has
this purpose and try to understand (1) whether an increase of incentive beyond a threshold level bring
worse performance; (2) examine the generality of any detrimental effect of incentive.
EXPERIMENT 1:
In the first experiment 87 participants in rural India were recruited. Poor people (low education,
someone had a bicycle or a TV). The experiment was conducted testing one participant at a time and
these people were randomly assigned to one of the three treatments: small (4 Rs), moderate (40 Rs)
and very large incentives (400 Rs, about 80% of the average monthly consumption expenditure). In
each treatment, participants played six different games divided in three categories: creativity (e.g.,
Packing quarters), memory (Simon), motor skills (Labyrinth) and the payment were given only if they
reached certain levels, namely “good” (50% of incentive above) and “very good” (100% of incentive
above). From this experiment we can see that: (1) there are no differences between low- and mid-
payment conditions probably because the incentive in the low- condition already maximized the
performance; (2) in the high payment, performance was always low compared to the other two
treatment; (3) There were no differences about memory tasks and the other two categories. This is
probably due to the fact that the memory tasks involved some skills susceptible to some choking-
generating mechanism. Another problem of this experiment could be that the incentives chosen were
too high, hiding in this way the differences between the games. Another limitation could involve the
participants of the experiment: poor people from India, unfamiliar with games and very different from
the people who leave in the advanced economies.
EXPERIMENT 2:
The second experiment tries to fill some shortcomings of the first. It used different participants: 24 MIT
undergraduate, using familiar tasks and using within-subject design in which each participant received
high and low treatment. There were two tasks: (1) Key pressing: required only physical effort; (2)
Adding numbers: required cognitive effort. This experiment helps us to examinate two postulates: (1)
High performance-contingent incentives increase pure effort and, by doing so, also the performance
based on pure effort; (2) High performance-contingent incentives can decrease performance based on
cognitive skills. So, we expect to see an increase of performance in “Key pressing” and a performance
decrease in “Adding”. About treatments: (1) Low incentive: 15 USD for “good”, 30 USD for “very good”
performance; (2) High incentive: 150 USD for “good”;300 USD for “very good” performance.
As the scholars supposed, in the “key pressing” tasks increase as the incentives increases. While in the
“Adding” tasks, performance decreased as a function of stakes: that means that there is a level of
incentive beyond which further increase lead to a decrease of performance.
EXPERIMENT 3:
The 3rd experiment examines the social incentives, in particular: (1) examine the social incentives in
the contest of financial incentives and (2) it investigates possible gender difference (see also Gneezy,
Niederle and Rustichini (2003)). As Zajonc (1965) said in his “social facilitation” framework, audience or
coaction increases arousal and it could be positive or negative. Follow-up papers show a more
cognitive model in which this effect depends more on individuals’ perception of the situation and their
personality. Charness, Rigotti and Rustichini (2007), show that the behaviour of people is affected by
audience in games.
The experiment took about 30 minutes: in 5 session at University of Chicago, for each session there
were 7/8 participants and they had to try to solve anagrams, 24/26 one-minutes trial. Sessions were
private or public and each participant played 10 private and 2 public games. Payment was the same for
public and private trials. Results show, as expected, better results in a private condition than in public
one. This because (1) anagrams involve creativity and (2) audience produces high level of motivation
that lead to choking under pressure. No difference has been found between gender neither in solving
anagrams, nor in being influenced by the social pressure.
From these experiments we can affirm that not always an increase in motivation lead to improvements
in performance and the fact that there were registered better performances with low incentives than
with high incentives is coherent with the “Yerkes-Dodson law”: an increase beyond a threshold level is
detriment for the performance. The optimal level of arousal is not always the same but can vary with
the task, the individual’s personality, and experience with the task. For more practical tasks, for
example, this level could be higher. These experiments found no proof a favour of the idea that there
are systematic individual differences in the propensity of choke (only in the experiment 2 they
revealed a negative rather than a positive correlation probably because higher incentives help in the
Key pressing task but disturb the Adding task).
Regarding the 1st experiment: it could be interesting to see the response of the people doing the
games under the three incentive conditions. The 2nd experiment, instead, suggests to do more
accurate research about the relation between choking under pressure, individual characteristics, and
task-specific characteristics. The relation between the 3rd experiment and the findings by Falk and
Ichino (2006), suggest that there is a relation among social incentives, performance, and type of
payment.
To sum up, since we have seen that stronger incentives do not necessarily lead to better performance
(especially in tasks involving cognitive (difficult) effort), it would be useful to further and better
understand this topic because it could be help improving the system of incentives with the many
positive effects that can ensue like: saving money, increment the marginal productivity of labourer
and, consequently, better profits.
REFERENCES
ARIELY, A., GNEEZY U., LOEWENSTEIN G., and MAZAR N. (2008), “Large Stakes and Big Mistakes”,
Review of Economic Studies (2009) 76, 451–469
