Wednesday, December 5, 2018

When fractions are better than decimals


In order to graduate from high school, people usually spend  about13 years (K-12) learning about math concepts.  One of the things that math does is to give us a universal way to think about quantities.  The beauty of the number 2, for example, is that it refers to a pair of items regardless of whether those items are bowling balls, butterflies, or beer barrels.
Even though math is about abstract concepts, the human mind is often focused on specific situations in the world.  As a result, the mathematical notations we learn may not always make it easy for us to reason about the world.  Psychologists have begun to explore the relationship between the way people naturally want to reason and ways that we can represent situations using numbers.
An interesting paper in the February, 2015 issue of the Journal of Experimental Psychology: General by Melissa DeWolf, Miriam Bassok, and Keith Holyoak explored differences in the way that fractions and decimals affect thinking. 
A fraction takes two integers and places them in a ratio (like 3/4 or 10/15).  A decimal can express the same numerical quantity, but it does so with a single number (like 0.75 or 0.66).  So, the fraction makes the relationship among numbers clear in the way it is written, while the decimal does not.
In one study, the researchers showed college students displays in which a fraction or decimal could be used as a description.  In the continuous displays, there was a rectangle and some of it was shaded red, while the rest was shaded green.  In the discrete displays, there were several objects, some of which were red, and some of which were green.  Finally, in the discretized displays, there was a rectangle (like in the continuous display), but it was divided into regions of equal size.  Some of those regions were red, and some were green.
With displays like this, there are two kinds of comparisons people can make.  There are part-to-whole comparisons. For example, if there are three red squares and five green squares, then the part-to-whole relationship of red squares is 3/8.  There are also part-to-part comparisons.  In this same display, the relationship between red squares and green squares is 3/5. 
In one study, participants were shown a continuous, discrete, or discretized display and were asked whether they would prefer to describe either a part-to-whole or part-to-part relationship for that display with a fraction or a decimal.  They weren’t asked which fraction or decimal they would use for the displays, just whether a fraction or a decimal would feel more appropriate.  For both kinds of relationships, participants preferred to use decimals for continuous displays, and to use fractions for discrete and discretized displays.  A second study asked people to identify the specific relationship shown in a display and found that people were equally good at using fractions and decimals for continuous displays, but much better at using fractions than decimals for discrete and discretized displays.
So far, these results are pretty straightforward.  The rest of the studies explored the ability to perform mathematical analogies.  In these studies, participants saw a display and either a fraction or a decimal that described the part-to-part or part-to-whole relationship in that display.  For example, if they saw 3 red squares and 5 green ones, and the fraction 3/5, that was describing the part-to-part relationship.
Next, they saw a second display of the same type, and two descriptions of the relationship.  For example, this time they might see 5 red stars and 7 green stars.  They would see one mathematical description of the part-to-part (5/7) relationship and one description of the part-to-whole relationship (5/12), and they had to pick the response that referred to the same relationship that was shown in the first display.  In this case, they would have to pick the part-to-part display.
They found that people had a hard time with this task for both fractions and decimals when the displays were continuous.  They were much better at the discrete and discretized analogies with fractions than with decimals.  This was true, even when the fractions referred to the same ratio, but did not map directly onto the number of items in the display.  For example, if the part-to-part relationship in the display with 3 red squares and five green squares was described with the fraction 6/10, that would be equivalent to 3/5, but the numbers would not map directly to the display. 
This work fits with a lot of previous research suggesting that people like to reason about frequencies of things in the world rather than proportions.  We experience the world in terms of the numbers of objects we see and the numbers of events we experience.  Math allows us to create other representations like decimals that are great for calculations, but they can make it harder to reason about what has happened in the world. 
This work also suggests that if you are staring at a numerical description of a situation and that description does not make sense to you, consider trying another way to think about it.  You are often asked to make decisions based on information that involves numbers.  Often, those numbers are decimals or proportions.  Consider turning those numbers into frequencies or fractions when reasoning about them.

Thursday, November 29, 2018

Peer Pressure Affects Your Actions More Than Your Recommendations


As a parent, I am often confronted with the hypocrisy of advice-giving.  There are plenty of things I have recommended to my kids that are courses of action that I have not taken myself.  Some of that is that I want my kids to avoid some of the mistakes I have made.  But, some of it is also that the way you give advice differs from the way you decide what to do yourself.
An interesting paper in the February, 2015 issue of the Journal of Experimental Psychology: General by Sarah Helfinstein, Jeanette Mumford, and Russ Poldrack examined an important factor that leads people’s recommendations for others to differ from what they themselves would do. 
Participants were asked about a variety of risky behaviors across a number of domains including social risks (like moving away from family), recreational risks (like bungee jumping), financial risks (like betting a day’s wages at poker), safety risks (like driving without a seatbelt) and ethical risks (like not returning a wallet with a lot money in it).  These were taken from a normed inventory called the DOSPERT.
Participants all rated the potential benefit, potential cost, and likelihood that they would incur the cost for each of these risky behaviors.  They also rated how likely it was that other people engaged in these behaviors.  Finally, participants either rated their willingness to engage in that behavior themselves or their willingness to recommend the behavior to someone else.
Overall, people were not really more or less willing to engage in risky behaviors themselves than to recommend them to others.  In some domains (like social and safety risks, people were more willing to perform the actions themselves than they were to recommend them to others.  In other domains (like recreational and financial risks) people were less wiling to perform the actions themselves than to recommend them to others.
It is the determinants of these recommendations that are most interesting.  People were more likely to perform an action and to recommend it when there was perceived benefit for doing it and less likely to perform or recommend it when there was a significant cost and when that cost was seen as being likely to happen. 
The big place where willingness to perform an action differed from the recommendations people made was in the influence of the likelihood that other people perform the action.  When recommending an action to others, the likelihood that other people perform the action did not matter much.  But, engagement of other people increased people’s willingness to perform the action themselves.  That is, people succumb to peer pressure.  When we perceive that other people are performing an action, it makes us more likely to perform it ourselves. 
This tendency is true for decisions that do not involve risk as well.  For example, studies of consumer products demonstrate that the leading brands in different product categories has remained relatively stable for a long time.  Gillette has been the leading brand of razors and Tide has been the leading brand of detergent for over 50 years.  Part of what helps brands like these to maintain their dominance is their perceived popularity.  Even without knowing it explicitly, we buy what we think other people are buying. 
When you are contemplating a decision, then, it is worth asking yourself both what you are thinking of doing as well as what you might recommend to someone else.  If you think that there is a difference between what you would do yourself and what you would recommend, then stop and think about whether you might be better off doing what you would recommend to someone else. 

Friday, September 21, 2018

Can Liars Be Caught By Aviation Security?


If you travel frequently, then you have probably endured more than one security screening interview at an airport.  At passport control, for example, border agents ask a few questions, stare at your passport, check you on electronic databases, and then send you on your way.
The purpose of these interviews, of course, is to catch people who are being deceptive.  This is a remarkably difficult job.  Thousands of people are streaming through airports every day, and only a very small fraction of them are actually being deceptive in any significant way. 
Can those few liars be caught effectively?
This question was explored in a field study by Thomas Ormerod and Coral Dando published in the February, 2015 issue of the Journal of Experimental Psychology: General. 
They trained two samples of security agents on different methods of screening for deception. 
One method focused on finding suspicious signs that are supposed to distinguish lairs from truth-tellers.  This technique, which is typical of what is taught to security screeners, involves a scripted set of questions.  As passengers answer these questions, agents look for signs of deception like nervousness or a passenger dressed inappropriately for the trip they are going on.
The other method focused on the security interview itself.  This technique uses a lot of open-ended questions designed to get the passengers to talk about themselves.  The agent asks questions of general knowledge that someone with that background should know.  For example, a passenger who claims to be a student at the University of Texas should know where the Student Union is or how to get to the airport from campus.  The technique is designed to give the passenger very little control over the direction of the interview.  Deceptive individuals like to control the conversation, so that they can focus on the details that they want to discuss.
After this training, about 200 deceptive passengers were sent through security over an 8-month period.  Notice, this means that most passengers that any agent saw were genuine passengers.  The deceptive passengers were recruited to participate in the study.  They were given payment to be in the study and a significant amount of additional money if they passed through security without being called aside for further screening.  So, the deceptive passengers had some incentive to get through security.  The ability to detect whether this group was deceptive was compared to the screeners’ likelihood of sending genuine passengers for extra screening whose demographic characteristics matched those of the experimental sample. 
A few weeks before participating in the study, the deceptive passengers were given a cover story to learn that was tailored to their age and appearance.  The example given in the paper was a male police officer who was told to imagine that he was working as an engineer near his hometown and was traveling to Chicago to take part in a fencing competition.  The deceptive passengers were told to do research to add information to their backstory.  A subset of the deceptive passengers were also given several opportunities to be deceptive at different times, to see if they could improve in their ability to fool the screeners.
The results were striking.  Screeners trained with the interview technique caught about 70% of the deceptive passengers, while those trained with the traditional method of looking for cues detected about 5% of the passengers.
You might thing this reflected that the screeners trained with the interview technique just sent more people for additional questioning than those trained in the traditional way.  That was not the case.  The agents who learned the interview technique were no more likely to send other passengers for additional screening than those who were trained in the traditional way. 
You might also think that the interview technique takes longer.  It does not.  Agents trained with each technique spent about 3 minutes with each passenger.  The big difference was that those trained with the interview technique asked more open-ended questions and gave the passengers more time to speak than those trained in the traditional way.
Finally, those passengers who were given several opportunities to get through screening were caught at about the same rate for each try.  This finding suggests that it was not straightforward for passengers to learn to beat the system. 
The reason why the system is hard to beat is straightforward.  If you claim to be someone very different from who you are, there is a wealth of life experience that you simply don’t have.  All of the specific details of life from where you shop to how you drive home to what buildings you pass on your way home are second-nature when you are telling the truth and are absent when you are involved in a large-scale deception.  These security interviews allow agents to capitalize on the absence of this knowledge when interviewing passengers.
Finally, this research is a great demonstration of the way that psychological research can be used to solve a practical problem.  First, the interview technique itself is drawn from extensive research on deception.  Second, the test itself is quite well-constructed, and the researchers do a good job of ruling out a number of alternative explanations for the results.

Wednesday, August 29, 2018

Conflicting Goals Can Make You A Better Decision Maker


We tend to think of conflict as the enemy of good decision making.  We dread situations that involve difficult choices.  Indeed, studies by Amos Tversky, Eldar Shafir, Ravi Dhar, Itamar Simonson and their colleagues suggests that people will actually avoid making decisions that are difficult.  When given a choice between selecting one of two options that require making a difficult tradeoff (for example, selecting apartments that differ in size and commute time), people prefer to put the decision off until later rather than addressing it right away.
An interesting paper by Jennifer Savary, Tali Kleiman, Ran Hassin, and Ravi Dhar in the February, 2015 issue of the Journal of Experimental Psychology: General suggests that there may also be an upside to experiencing conflict. Specifically, they suggest that when people have two conflicting goals that they are grappling with, that makes them likely to think carefully about choices in order to resolve the conflict.
In order to induce conflicting goals, participants did a lexical decision task in which they saw a series of letters and had to press one button of those letters formed a word and a second button if they did not form a word.  In the conflict condition, some of the words referred to a particular goal (such as being healthy, with words like fitness and active), and others referred to a second goal that conflicts with the first (such as indulgence, with words like decadent and indulge).  The control condition did not have conflicting goals embedded in the words that were part of the lexical decision task.  Tasks like this have been used in many previous studies to activate goals and to create goal conflict.
After this lexical decision task, one study gave people difficult choices (like a choice between apartments that differ in size and commute time) and asked people to select one of the options or to defer the choice until later.  Participants who were induced to feel a conflict between goals were actually more likely to choose one of the options rather than deferring the choice than people in the control condition who were not given a goal conflict.
In a second study, participants were given these choices using a computer system that tracked the amount of time participants spent making the decision and the number of features of the options they explored.  Participants induced to experience a conflict looked at more features and spent more time making the choices than those who did not experience a conflict.  This study also demonstrated that people were not aware of the goal conflict that was induced. 
One other study tested the idea that conflicting goals increase how thoroughly people process information about choices in a slightly different way.  Again, goal conflict was induced using the lexical decision task.  This time, though. The decision task involves selecting from among three options (say three different apartments).  One was very good on one dimension (it was large), but very bad on the other (it was far from work).  A second was bad on that first dimension (it was small), but good on the other (it was close to work).  A third was a compromise (medium in size, a moderate commute to work). 
Previous research suggest that when people don’t want to work that hard making a choice, they tend to select the compromise option so that they don’t need to figure out which dimension is more important to them.  If people really think carefully about the choice, then, they will be more likely to pick one of the extreme options rather than the compromise. 
Consistent with the other two studies, participants induced to have a goal conflict were more likely to pick one of the extreme options than people in the control condition who had no goal conflict. 
An interesting aspect of these studies is that the goal conflict that was induced was not directly related to the choices people were making.  So, the increase in depth of thought about the choices was caused by the presence of active goals that conflict, and not based on the activity of goals that were relevant to evaluating the options.
This research suggests that we experience two kinds of conflicts when making choices.  One conflict is between options that are about equally attractive and require tradeoffs among the features to figure out which is best.  These conflicts make it hard for people to choose.  Often, people prefer to defer the choice until later or pick an easy compromise option rather than resolving tradeoffs.
The second kind of conflict is one between incompatible goals.  These goal conflicts arouse the motivational system.  This arousal leads people to consider options more carefully, think about them more deeply, and ultimately helps people to make the tradeoffs that can make decisions difficult.

Monday, August 6, 2018

Sleep, Thinking, and Aging


I have written a few times about the influence of sleep on thinking.  High school students who stay up late perform more poorly in school the following day.  A lack of sleep may cause you to mix together different memories that did not occur together.  In young adults, sleep also affects the ability to learn new procedures. 
These benefits of sleep lead naturally to the speculation that sleep may help older adults avoid the cognitive declines that come along with aging.  One possibility is that older adults who suffer from sleep difficulties decline faster than those who don’t.  Another possibility is that regular sleep throughout life is associated with lower levels of problems.
A paper in the January, 2015 issue of Perspectives on Psychological Science by Michael Scullin and Donald Bliwise tried to sort out what is going on with sleep and aging.  The performed a massive meta-analysis.  A meta-analysis looks across the many published studies in an area of research in order to explore what really seems to be happening in an area.
There are many ways to study sleep and its effects on thought and aging.  Some studies use self-reports of sleep quality and measurements of cognitive performance.  Some of these self-report studies look at people of different ages.  Others are actually longitudinal.  They examine the relationship between the quality of sleep people get at one point in time and their performance later in their life.
Other studies use other measures of sleep.  Some studies use a device called an actigraph, which measures whether the person is moving.  (The Fitbit is a kind of actigraph.)  Long periods without movement are good (though not perfect) signals that a person is sleeping.  Still other studies measure physiological aspects of individuals like brain waves so that it is possible to tell both that people are asleep as well as the stage of sleep they are in.  Finally, there are experimental manipulations of sleep including sleep deprivation studies as well as studies in which people are randomly assigned to conditions in which they do or do not nap.
There are a lot of interesting findings in this paper, and it is worth giving it a read yourself for a more complete look at effects of sleep on thinking.  But here are a few highlights.
First, the relationship between sleep and improved thinking is strongest earlier in life and gets weaker later.  A good night’s sleep helps young adults to learn better the next day.  Sleep also helps young adults to consolidate (or solidify) memories from the day before more than it helps older adults.  Middle-aged adults show smaller effects of sleep on learning, and older adults show almost no relationship between sleep and learning at all. 
Sleep deprivation studies tell the same story.  Sleep deprivation generally hurts thinking performance, but these effects are much stronger in younger adults and small or even non-existent in older adults.  (This may explain why I can play the sax in a blues band until 2am on Sunday nights and still function at work the next day.)
Of course, part of the difficulty with studying sleep in older adults is that older adults generally need less sleep than younger adults, and the older adults who get the most sleep tend to be those who are sick and whose bodies are fighting off illness.
These results do suggest, though, that the amount of sleep that older adults are getting at that phase of their lives is not a cause of cognitive decline.
A particularly interesting result is that the quality of sleep in middle age influences cognitive health in old age.  The longitudinal studies are particularly helpful for this work.  When adults in their 40s and 50s get regular sleep and allow themselves to get the roughly 8 hours of sleep they need, they show fewer signs of cognitive problems like senile dementia when they are older.  Indeed, one of the studies in this sample measured sleep quality of adults in their 40s and followed up with them 28 years later.
Putting all of this together, then, it seems that sleep is most important for current cognitive performance in younger people, and that sleep plays less of a role in thinking as we age.  Sleep in middle-aged adults is still important, though, because good sleep habits in middle-age are associated with better mental health in old-age.

Thursday, June 28, 2018

What Does Your Avatar Say About You?


A lot of websites give you the chance to represent yourself with an avatar rather than a picture of yourself.  Avatars are often cartoon-y pictures with facial features, clothing, and accessories that allow you to personalize your picture.  For example, this website allows you to create an avatar to use before entering a chat room.
The avatar you select can influence the way people interact with you.  It is interesting to know whether people generally try to select avatars that represent themselves accurately, or whether they aim to display themselves differently to the electronic world than they appear in real life.  It is also interesting to know the conclusions that viewers draw when seeing someone’s avatar.
This question was addressed in a study by Katrina Fong and Raymond Mar published in the February, 2015 issue of Personality and Social Psychology Bulletin. 
The researchers asked a group of about 100 people to choose avatars for themselves using the (now defunct) website weeworld.com.  Half of the participants were asked to create an avatar, and the other half were specifically asked to create an avatar that would represent their personality accurately.  There were no significant differences in the avatars created by these groups suggesting that most people naturally try to represent themselves accurately.  These participants filled out a personality inventory that measures the Big Five personality traits after creating their avatar.  (The Big Five traits are Openness, Extraversion, Agreeableness, Conscientiousness, and Neuroticism.)
A second group of about 2,000 participants were shown a subset of the avatars and rated their perception of the personality characteristics of the individuals who created those avatars.  They also rated how much they would like to interact with the person who created that avatar. 
One question that the researchers asked up front was whether being able to categorize the participant by gender influenced judgments of personality.  The avatars were all either recognizably male or female.  Overall, people tended to think that the males were slightly less conscientious and open to new experiences than the females.  But, this categorization tended to decrease accuracy of judgments overall, because the sample of male participants was not actually lower in conscientiousness or openness than the sample of female participants.
The researchers compared people’s ratings of their own personality characteristics to those of other people who rated personality after seeing the avatars they constructed.  The ratings of the avatars showed that people could assess another person’s extraversion and agreeableness to some degree, and could not do a particularly good job of rating the other characteristics.
The researchers also examined the aspects of the avatars that were most correlated with people’s personality ratings.  For example, people high in agreeableness tended to select avatars with open eyes more often than those low in agreeableness.  One reason why raters were good at assessing an individual’s agreeableness from their avatar was that they generally rated people as higher in agreeableness (and extraversion) if the avatar had open eyes. 
In general, though, the aspects of avatars that raters thought were most important for judging a person’s personality were not that diagnostic of the personality characteristic.  For example, people tended to rate avatars with short hair as more conscientious than those with long hair.  In fact, this characteristic was more strongly associated with the neuroticism of the person who created the avatar than the conscientiousness of that individual.  People higher in neuroticism tended to have avatars with long hair. 
One final data point of interest, the characteristics of avatars did influence whether people were interested in befriending the person.  In particular, people were most interested in being friends with people who had avatars with open eyes, smiles, and an oval face and were least interested in being friends with people who had a facial expression that was not a smile.
So, what does all of this mean?
There has been a lot of work recently on what we can learn about the personality characteristics of others from the things they create including personal spaces, Facebook pages, and things they write.  Overall, when people create an avatar, it is hard to get to know much about them.  You can get a little information about extraversion and agreeableness, but the correlations are not large. 
One thing that is interesting, though, is that people do draw inferences about personality characteristics from avatars.  However, the aspects of the avatars that they use to make judgments about someone’s personality are not generally that highly correlated with that individual’s actual personality.  Thus, people may overestimate their ability to learn something about others from their avatars. 

Thursday, June 21, 2018

Why Do Movies Move?


If you spend time watching movies or TV, you have probably know that you see a moving image on the screen, but that the sense of motion is created by your brain from a series of static images.  Typical movies, for example, flash 24 frames per second.  Somehow, the brain takes the changes from one frame to the next and gives you the illusion of fluid movement. 
How does that happen?  Take a moment to try to explain it to yourself.
This question is just one of many that is explored in a great book that came out in 2015 called Flicker by cognitive neuroscientist Jeff Zacks.  The book itself explores a variety of topics ranging from low-level aspects of the way the visual system understands the images on the screen all the way to high-level topics like the reason why movies are so good at creating emotion.
So, why do movies move?
You might think that what is happening is that each image persists a little on the retina (the cells at the back of your eye that respond to light) and that changes in the image are detected there.  Or perhaps, when the image is first processed in the brain, it recognizes small discrepancies from one frame to the next.
Neither of those possibilities is quite right.
The images from the screen enter the eye and hit the retina.  From there, they are passed into the brain and ultimately make their way to the occipital lobe in the back of the brain where most visual processing is done.  Initially, the brain looks for simple visual features in the image like the presence of edges, because edges usually signal the boundaries of objects.
The interesting thing is, the brain divides up the task of understanding the image in multiple ways, with different brain areas searching for different features in the image.  The unified sense of vision we have arises because the brain ultimately puts all of those independent properties back together. 
Early in the processing of images a particular area of the brain called area MT (shown in the figure) looks for blobs that have changed position.  When MT sees a blob in a location that has changed it position a bit, it gives a signal suggesting that there was motion.  Sustained activity of MT indicating motion in a particular direction gives people the experience that an object moved.  This brain area doesn’t really care much about the blobs themselves.  The blobs could actually change shape or color from one frame to the next, it is just looking for motion.
So, the motion in movies comes from activity in the brain area MT.  As Zacks points out, though, this can sometimes cause problems.  In particular, when a movie is put together, it is usually constructed from a set of scenes that are spliced (or cut) together.  If the editor is not careful, when one scene is cut to the next, some of the objects may appear to jump from one location on the screen to another.  This jarring sense of movement is called a jump cut.  Filmmakers try not to create these jump cuts, and texts on film making give suggestions for how to avoid them. 
These jump cuts are caused by the same process that causes the sense of motion in scenes.  When one scene is cut to another, if area MT detects the motion of a blob, it will send a signal that an object in one location actually moved to another.  That can feel weird, because brain areas that calculate the size, color, and shape of the object may not see strong similarities in the objects from one scene to the next, so you can get a feeling of motion without have a clear sense of what moved.
One of the reasons why Flicker is an interesting read is that Zacks explores the ways that movies exploit the structure of the brain to give us an immersive experience.  If you can pull yourself away from the screen for a few hours, it is well-worth the time to check it out.