Learning by Accident: Some Memory Mistakes Are Not Totally Wrong

When your parent picks you up from school, how do you know which car is theirs? You probably “just know,” even without effortfully trying to memorize the brand, type, or color of the car. Although we often think of memory in terms of intentionally studying for tests, many memories are formed accidentally, without you even trying to intentionally learn something. But how detailed are these accidental memories? Do pieces of information remain, even if we feel like we have forgotten them? To explore these questions, we conducted an experiment in which participants searched through pictures, looking for specific objects. Later, and without warning, we asked people to choose previously seen pictures from a group of similar objects. Even when participants’ choices were wrong, they often chose pictures that looked much like a picture they saw earlier. These “mistakes” do not reflect total forgetting, but instead show that people remember more than they think!


CAN WE REMEMBER THINGS "BY ACCIDENT"?
Our visual world is gigantic; thousands of shape and color combinations make up all the distinct scenes and objects we see every day. During commercial breaks when watching TV, for instance, you probably do not care about the visuals in the ads: you see them, but you do not really pay close attention to them. Yet, your ability to remember specific pictures is much better than you might think. For example, research has shown that people can learn and recognize close to , di erent pictures, even if they only study each for a few seconds! [ ]. And what is even more surprising is that many of these visual memories are incidental memories, that is, they are

INCIDENTAL MEMORY
Information learned "by accident," without trying to remember what is in front of you. created almost by accident, without people even trying to remember what they saw.
We typically think of learning and memory as being intentional (on purpose), like sitting in a classroom trying to pay attention so you can remember everything later, during the test. Some memories, particularly for school-related information, are formed this way, and we retain the knowledge while gradually forgetting the details of when and where we first learned it. For example, you probably do not remember much about the place or day when you first learned the alphabet. However, many of our visual memories, such as those for places and the objects within them, are actually formed incidentally. Even though we do not try to remember these things, we can retain an impressive amount of detail. For instance, imagine that you are in the lunchroom looking for your friend Lily, who is saving you a seat. As you scan the room, you see people sitting at di erent tables and wearing many di erent color clothes, including Je , who is wearing a brown jacket. You eventually find Lily, have lunch, and head back to class. If, later that day, Je tells you that he lost his jacket, would you be able to help him pick it out of the cluttered lost-and-found box? In this example, you were not paying close attention to Je (you were looking for Lily), and probably only saw his jacket for a few seconds. Despite this, research shows that you would correctly spot Je 's jacket in tries out of [ ]. Some studies have even shown that you learn and remember visual information equally well whether you are trying to or not! [ ].

MEMORY "MISTAKES" ARE VERY INFORMATIVE
Although many studies demonstrate that you can learn lots of visual information without trying [ -], what would happen if several other jackets in the lost-and-found were also brown? Are incidental memories detailed enough that you would still find Je 's jacket? Assessing the quality of memories is not easy. When researchers "measure" memory, we normally focus on a person's ability to choose the single correct answer in a memory test. If three of the jackets in the lost-and-found are brown, yellow, and blue, recognizing Je 's brown kids.frontiersin.org September | Volume | Article |

Figure Figure
A visual search task in which participants searched for one to four potential targets on each try, not knowing which, nor how many, of the target pictures would appear in the display. In the actual experiment, we used color images of real-world objects instead of clipart drawings.
jacket is easy. If, however, there are many lost jackets and a few of them are brown, you might end up bringing Je the wrong brown jacket. Although this is a mistake, you still remembered enough to choose a similar jacket.
Now imagine that there are jackets, and they are all brown! As the di erent options become more similar, recognizing the correct choice becomes more di cult, and memory mistakes become more common. Contrary to popular belief, these mistakes do not mean you have no memory for the object you are trying to remember. If you had no memory, you would be equally likely to bring Je a blue jacket as a brown one, as if you were randomly guessing. Instead,

RANDOM GUESSING
Equal likelihood of choosing any of the available options. In a coin toss, for example, your chance of randomly picking any one of the two outcomes is % ( out of ). mistakenly choosing a similar brown jacket indicates that you do have a pretty detailed memory for Je 's jacket, even if that memory is not "perfect."

OUR INCIDENTAL MEMORY EXPERIMENT
To measure how well-objects are remembered incidentally, we conducted an experiment in which people searched through pictures. This was similar to Where is Waldo? ® , except that people had to quickly find pictures of specific objects among pictures of other objects. For example, how quickly do you think you would have found one of the "target" pictures in Figure ? People had s to use the computer mouse to click on the target picture as quickly as they could. This means that they had to quickly find the target picture and try to ignore the other "distractor" pictures. The participants looked for di erent target pictures among new sets of distractor pictures over and over again− times! After all searches, we surprised them with a memory test for the distractor Figure   Figure Experiment participants were given a surprise memory test after the experiment explained in Figure . In this example, participants saw one of these backpacks as a distractor during the visual search task. In the memory test, they had to choose which of the four similar-looking backpacks they saw. As in the visual search task, colored pictures of real-world objects were used in our actual experiments.
pictures that they tried to ignore earlier. For instance, without looking back, which backpack in Figure appeared in Figure ? If you chose "d" then you chose correctly! On average, people in our experiment chose the correct answer over % of the time (around out of attempts). This might not sound like a lot to you, but if people randomly guessed by closing their eyes and picking, they would have only been correct % of the time (about out of attempts). Remember, this is incidental memory: in their searches, people tried to ignore, not remember, these pictures.

HOW ACCURATE IS INCIDENTAL MEMORY?
Knowing that people incidentally remember over % of the objects does not tell us how well-they remembered them. To answer this question, we looked at the mistakes people made. If you look back at Figure , you probably notice that option "c" does not look at all like the correct answer. On the other hand, option "b" looks very similar to the correct backpack, and option "a" is somewhere in the middle. This is the sort of multiple-choice test people in our experiment took-they had to choose the correct object from others that were "very similar," "moderately similar," and "not similar" to the correct answer.
Although the concept of similarity is intuitive, we used similarity maps

SIMILARITY MAP
A visual representation of how alike di erent items seem to people. The closer items are located to each other in the map, the more alike people think they are. to scientifically measure how much pictures did (or did not) look alike [ ]. Similarity maps can be created by asking people to arrange pictures according to how similar they think they are ( Figure A). Similar pictures are put close to each other, and as pictures become more and more di erent, they are located further away from each other. When lots of people do this task, researchers can then get reliable similarity maps for many categories of objects.
Using similarity maps, we found that the "very similar" (but incorrect) choice was selected more often than the "moderately similar" and "not similar" choices ( Figure B). This was not random guessing! If people guessed randomly, they would have been equally likely to pick each object. Instead, they selected the "very similar" object most often and kids.frontiersin.org September | Volume | Article | Guevara Pinto et al. the "not similar" object least often. This is like bringing Je the wrong brown jacket instead of bringing him the wrong blue jacket. Both jackets are "wrong," but one is less wrong than the other. Importantly, this suggests that visual items are not entirely forgotten: some part of the memory persists, even if we answer incorrectly.

Learning by Accident
Although the search example shown in Figure  is pretty easy, we increased the di culty in some aspects of our experiment. For example, sometimes people only searched for one thing (like a jacket). Other times, they searched for three things (like a jacket, a dog, and a rocket) at the same time, but only one of those things might appear, so people never knew what to expect. When the task became more challenging like this, people actually had better memory for the distracting pictures! We believe this occurs because people must pay close attention to the pictures when they are less sure of what they might (or might not) find.

CONCLUSION
Our experiment demonstrates that picture memories, including those formed incidentally, are quite detailed, even when we cannot perfectly remember what we saw. Perfect remembering is actually rather rare, especially if you do not pay attention and study hard! However, having less-than-perfect memories does not mean that we have totally forgotten. Instead, we can forget some aspects of a picture while remembering others, which shows that the memory is in your brain somewhere. Some researchers think that we store individual pieces or features in memory, and then we later "remember" by putting the whole picture back together. Indeed, when you try to remember a picture, the same regions of the brain's visual cortex become active as if you are actually seeing the picture! . . Hout, M. C., Goldinger, S. D., and Brady, K. J.
. MM-MDS: a multidimensional scaling database with similarity ratings for object categories from the massive memory picture database. PLoS ONE. :e . doi: . /journal. pone.