Social Cognition

Mini Review

• The nine effects so far are about judging a description, another person, yourself, other people’s minds, and your own knowledge.
• Each one is a judgment made with the information that is in mind at that moment.

Hindsight Bias

• Hindsight bias: The tendency, once an outcome is known, to believe that it was more predictable than it was, and to remember having expected it.
  • Ex: After a team loses in the playoffs, a fan says he knew they would lose, though he had bet on them the week before.
  • Ex: After a stock falls, a man says the warning signs were there, though he had not sold it.
• Roese and Vohs (2012) describe three parts: the memory of the earlier prediction moves toward the outcome, the outcome comes to seem like it had to happen, and the person believes they could have foreseen it.

The Studies Behind Hindsight Bias

• Fischhoff (1975): Everyone read the same short history of an 1814 war between British and Gurkha forces, which stopped before the ending. One group was told nothing more (control group) and rated how likely each of four endings was (British win, Gurkha win, stalemate with a peace settlement, stalemate without one). Each of the other groups was told that one of the four endings actually happened before rating how likely all four were.
  • Finding: Whichever ending a group was told, they rated it about twice as likely as the group told nothing did, and said they would have predicted it. (Four endings instead of two let the study identify the bias for several endings, so we know it was being told about it rather than the specific ending itself that caused the difference.)

The Studies Behind Hindsight Bias, cont.

• The bias also occurs for predictions a person made themselves. The first study compared different groups told different endings; this one compares each person’s own prediction with what that person later remembers predicting.
• Fischhoff and Beyth (1975): Two weeks before President Nixon’s 1972 trips to China and the Soviet Union, students wrote down a probability, 0 to 100, for each of fifteen things that might happen on the trips. Weeks to months after the trips, the same students were asked to write down the probabilities they had given before, without seeing them.
  • Finding: For things that had happened, the probabilities they remembered giving were higher than the ones they had written. For things that had not happened, they were lower. Their memory of their own prediction had moved toward what happened.

Why Hindsight Bias Happens

• Once the outcome is known, it is the most available fact about the event, and a person reconstructing an earlier prediction starts from it.
• An outcome gets connected to the causes that led to it, so it reads as the natural result. The paths that led elsewhere are not reconstructed.
• Believing the outcome was foreseeable feels better than believing it was luck, for a win and for a loss.
• The effect is smaller when a person is asked to list reasons the other outcomes could have happened.

Correcting for Hindsight Bias

• Arkes et al. (1988): Neuropsychologists read a patient’s case and were told which of four diagnoses had turned out to be correct, then rated how likely they would have been to make each diagnosis from the case alone.
  • Finding: Those told the outcome rated it far more likely than those not told. However, when they first had to write one reason that each of the other diagnoses could have been correct, the bias disappeared.
• Listing reasons for the other outcomes works because it rebuilds the paths that led elsewhere, which the known outcome had erased.

Correcting for Hindsight Bias, cont.

• Guilbault et al. (2004): Combined 95 studies of hindsight bias.
  • Finding: The bias appeared across nearly every kind of event and was small to medium in size, so the correction has to be applied on purpose each time.

Exam 1 Polls

Two questions about Exam 1:

1. How many days before Exam 1 did you plan to start studying?
2. How many days before Exam 1 did you actually start studying?

Click Here to Answer

Exam 1 Poll Results

The Planning Fallacy

• Planning fallacy: The tendency to predict that one’s own task will take less time than similar tasks have taken in the past.
  • Ex: A family expects a kitchen remodel to take six weeks, and it takes four months.
  • Ex: A city budgets two years for a new bridge, and it opens in five.

The Studies Behind the Planning Fallacy

• Buehler, Griffin, and Ross (1994): Honors students predicted the date they would finish their thesis, and researchers recorded the date they actually finished.
  • Finding: The students predicted about 34 days, and the theses took about 56 days. Fewer than a third finished by their own predicted date.

The Planning Fallacy Since Then

• Halkjelsvik and Jørgensen (2012): Reviewed the studies in which people predicted how long a task would take and the time was then measured, with tasks from folding origami to writing software.
  • Finding: Predictions were too short in most studies and most settings, and the shortfall grew with the size of the task. For very short tasks the pattern sometimes reversed, with predictions too long.
  • Finding: People’s own records of similar tasks improved predictions when they were brought to mind, which is the correction on the slide after next.

Why the Planning Fallacy Happens

• A person planning a task pictures the steps of this task going as planned, which is the only version that comes to mind.
• Past tasks that took longer than planned are remembered as having done so for reasons specific to each one, so they are not used as evidence about this task. Explaining Behavior showed the fundamental attribution error: other people’s behavior is explained by who they are, and one’s own by the situation. A past delay is one’s own behavior, so it is explained by that day’s situation and never by a pattern.

Correcting for the Planning Fallacy

• The correction that works in studies is to predict from how long similar tasks took in the past, rather than from the steps of this one. This is called taking the outside view.
  • In Buehler, Griffin, and Ross (1994), students told to recall similar past tasks and apply them to the new one predicted later dates, and observers who read a student’s plan predicted later dates than the student and came closer to the truth. ## Correcting for the Planning Fallacy, cont. {.sm}
• Kruger and Evans (2004): People estimated how long a task would take, such as holiday shopping or formatting a document, either right away or after first listing the parts of the task.
  • Finding: Listing the parts first produced longer estimates, and the longer estimates were closer to the time the task actually took.
• Asking someone who is not doing the task works for the same reason: an outsider has no plan to picture and predicts from the record.

Impact Bias

• Impact bias: The tendency to overestimate how strongly and for how long a future event will affect one’s feelings.
  • Ex: A man expects that a raise will make him happy for the rest of the year, and by the next month his mood is back to usual.
  • Ex: A fan expects that losing the championship game will stay with her for weeks, and by the weekend she is planning next season.

What Counts as Impact Bias

• The bias is the gap between what a person predicts they will feel at some point after the event and what they report feeling when that point comes (Wilson and Gilbert, 2005). The prediction is usually right about the direction, worse or better, but too large about the amount and how long it lasts.
  • Ex: A student predicts that a low exam grade will leave her feeling much worse than usual a week later. A week later she feels a little worse than usual. The forecast was too large, which is the bias, even though she still feels something.
  • Ex: A year later she says the grade turned out to be a good thing. That is a different process, explaining the event, and it is one of the reasons the feeling faded.

Where Impact Bias Has Been Found

• The bias has been found for good and bad events, for events days and years away, and in people forecasting their own lives and other people’s. It is smaller for events a person has been through before.

The Studies Behind Impact Bias

• Gilbert et al. (1998): Assistant professors predicted how happy they would be in the years after being granted or denied tenure. Former assistant professors who had been through the decision reported how happy they were.
  • Finding: The forecasters predicted a large and lasting difference. The people who had been denied tenure were about as happy as the people who had been granted it.
• Dunn, Wilson, and Gilbert (2003): First-year students predicted how happy they would be living in the dorms they most and least wanted, before a lottery assigned them.
  • Finding: They predicted a large difference. A year later, students in the wanted and unwanted dorms were equally happy.

Why Impact Bias Happens

• When a person imagines a future event, the event fills the whole picture. Everything else that will fill those days is left out of the forecast.
• After an event, people explain it to themselves: a loss becomes the referee’s fault, a win becomes what the team had earned. Once an event has an explanation it stops producing much feeling, and people leave this out when they forecast, so they expect the feeling to last.
• Both reasons make the forecast too large in both directions, for good events and for bad ones.

Correcting for Impact Bias

• Wilson et al. (2000): Football fans predicted how happy they would be in the days after their team won or lost. Half first filled in a diary of what they would be doing on those days.
  • Finding: Fans who wrote the diary predicted smaller and shorter effects, and their predictions were closer to how they felt. ## Correcting for Impact Bias, cont. {.sm}
• Gilbert et al. (2009): Women predicted how much they would enjoy a speed date either from the man’s profile and photo or from one report by a woman who had already dated him.
  • Finding: Those given the other woman’s report predicted better, and most still said they would rather have had the profile.
• Both corrections put back what the forecast leaves out: the ordinary days around the event, or the experience of someone who has been through it.