Particularly Exciting Experiments in Psychology

The expression "two heads are better than one" reflects the intuition that people working in groups are more likely to come to a correct decision than they would if working alone. This may be because the individual with the correct answer is able to convince other group members because their argument is the most sound. However, because accuracy is often correlated with confidence, it may be that the most confident group member exerts the strongest influence regardless of whether their answer is right or wrong, and it just so happens that the most confident person is usually right.

Trouche, Sander, and Mercier (2014, Journal of Experimental Psychology: General) (PDF, 121KB) teased apart argumentation and confidence to determine what drives group reasoning decisions. Participants were given a problem from the Cognitive Reflection Test and had to provide their answer, their confidence in their answer, and a justification for their answer. They were then presented with an argument for either the correct or incorrect answer in written format with no linguistic markers of confidence (e.g., "it is obvious that…"). After reading the argument, they were given the opportunity to change their mind about their own answer.

Participants were significantly more likely to change their answer when their answer was wrong and they were provided with an argument for the correct answer than vice versa, indicating that good arguments lead people to change their minds in the absence of confidence cues. There was no relationship between initial confidence in one's own answer and the likelihood of changing one's mind.

In a subsequent experiment, participants were given a reasoning task individually, and after providing their answer, confidence, and justification, discussed the problem in groups of 3–5 until they reached a consensus. All groups adopted the correct answer if at least one group member solved the problem correctly individually. This was true even when the group member with the correct answer was not the most confident in their answer, or was less confident than the average of his/her peers. Thus, argument quality trumps confidence in group reasoning decisions.

While we typically think of group decisions involving reasoning, groups can also be asked to make perceptual judgments, such as when a team of physicians makes a diagnosis based on an x-ray. In this real-world example, all individuals see all of the perceptual information. However, it may be that group perceptual decisions would be better if the information was distributed among group members, allowing each individual to dedicate maximum focus to a subset of the information.

Barr and Gold (2014, Journal of Experimental Psychology: Human Perception and Performance) (PDF, 72KB) tested this possibility in a task where participants had to determine the presence or absence of a luminance signal in visual noise. Groups of 2–4 participants were tested in either a "Full" condition, where the entire stimulus was presented to all group members, or a "Partial" condition, where each group member was shown a unique quadrant of the stimulus. Participants made individual decisions before discussing their decision in their group to reach a consensus.

Full groups outperformed Partial groups, and the difference between Full and Partial groups increased with each additional group member. This suggests that group decision making benefits from sharing redundant information across group members, consistent with known statistical benefits of signal averaging. Interestingly, participants who ultimately made decisions in groups (Partial or Full) performed better on their individual judgments than participants who only completed the task in isolation. Thus, social motivation may lead to improvements in individual performance for those who know they will later be working with others.

Together, these studies confirm the intuition that two heads are better than one, and, reassuringly, this effect is driven by sound argumentation, statistical averaging, and social motivation, not the influence of strong personalities in a group.

Related Reading

• Barber, Harris, and Rajaram (online first, Journal of Experimental Psychology: Learning, Memory, and Cognition)taudrxda found that multiple mechanisms underlie the collaborative inhibition effect, where a group of people working together recall more than they would individually, but less than their predicted potential.

• Trouche et al. measured both reasoning accuracy and confidence. Explain why confidence could be a confound in an experiment where only reasoning accuracy is measured.
• Barr & Gold asked participants to determine the presence or absence of a visual signal. Fill in the table below showing how trials in their experiment would be categorized to apply signal detection theory.

	Respond "present"	Respond "absent"
Signal present
Signal absent

• What is a heuristic?
• What are the advantages and disadvantages of using heuristics to make decisions?

Return to PeePs homepage