Recall vs recognition: two memory tests, and when the easy one fails

Recall and recognition are two ways to get information out of long-term memory. In recall, you produce the answer yourself, as in an essay test. In recognition, the answer is in front of you and you identify it, as in a multiple-choice test. Recognition usually gives the higher score. But the size of that advantage depends on the test, and in some conditions recall gives the higher score.

• Recall is retrieval without the answer in view. Recognition is a comparison between an item in front of you and your memory.
• Recall has three common forms: free recall, cued recall and serial recall.
• Relearning is a third measure of memory. It is the measure that Ebbinghaus used in 1885.
• Recognition is usually easier, because the test item is a copy of the item that you learned. The wrong answers on the test also control the difficulty.
• In a 1973 experiment, students recognised only 24% of a set of list words. With the correct cues, students recalled 63% of the list.

What recall and recognition mean

Retrieval is "the act of getting information out of memory storage and back into conscious awareness". That is the definition in the OpenStax psychology textbook. The same textbook names three ways to retrieve information from long-term memory: recall, recognition and relearning.

Recall "means you can access information without cues". You use recall in an essay test. Nothing on the page contains the answer, so you must produce it.

The textbook defines recognition in two sentences: "Recognition happens when you identify information that you have previously learned after encountering it again. It involves a process of comparison." You use recognition in a multiple-choice test. The textbook gives a second example. At a school reunion after 10 years, you cannot recall the names of all your classmates. But you recognise many of them from their yearbook photos.

Researchers also name the two types of test. In a textbook module on memory, Kathleen McDermott and Henry Roediger write that psychologists use "production tests (involving recall) or recognition tests (involving the selection of correct from incorrect information, e.g., a multiple-choice test)". The word production is a good label. In recall, you produce the answer. In recognition, you select it.

Free, cued and serial recall

The word recall alone frequently means free recall. But recall has three common forms.

• Free recall: you produce the items "in any order", in the words of McDermott and Roediger. The test gives you no hint.
• Cued recall: the test gives you a hint, or cue, for each item. In an example from McDermott and Roediger, the first names George Bernard are a cue for the surname Shaw. A fill-in-the-blank question is a cued recall test.
• Serial recall: the APA Dictionary of Psychology defines it as "recalling items in the order in which they were presented". The dictionary gives a telephone number as the example, because the digits must be in the correct sequence.

Relearning: the third measure

Relearning "involves learning information that you previously learned", according to OpenStax. If you learn the material faster the second time, a part of the first memory remained. The OpenStax example is a woman who learned Spanish at school. She starts a Spanish course 13 years later, and she is surprised by her speed.

Hermann Ebbinghaus used this measure in his 1885 book on memory. He recorded the time that he needed to learn a list of syllables. Later, he recorded the time that he needed to learn the same list again. He called the difference "the saving of work in the case of relearning". He wrote that this saving "is each time the measure for the amount remembered at the end of the interval". Our article on what Ebbinghaus measured with the forgetting curve explains his savings figures.

Why recognition is usually easier

McDermott and Roediger give the usual reason: "We usually think of recognition tests as being quite easy, because the cue for retrieval is a copy of the actual event that was presented for study." In a recall test, you must find the item in memory. In a recognition test, the item is on the page.

Lionel Standing measured the difference in a 1973 study of memory for pictures and words. In one experiment, each group of 10 people saw 200 items one time, for 5 seconds each. The people then had 40 minutes to write all the items that they could recall. After that, they did a forced-choice recognition test.

For printed words, the mean recall score was 24.5 words. From the recognition test, Standing estimated that 134.5 words were in memory. That estimate is 5.5 times the recall score. For pictures, the two figures were 51.6 and 175.

Standing wrote that this pattern was a known one: "It has often been found that recognition performance exceeds recall, for a particular type of material and a given number of alternatives in the recognition test, by some characteristic factor." He also summarised a 1967 study by Shepard. In that study, people saw 612 pictures one time, for about 6 seconds each. In tests with two alternatives, they selected the correct picture 98% of the time. The figure for single words was 90%, and the figure for short sentences was 88%.

The wrong answers control the difficulty

Standing included a condition in his sentence: "a given number of alternatives". A recognition test does not have a fixed difficulty. Standing made his own recognition task harder when he used more alternatives.

In 1983, J. F. Hall wrote a short paper about this problem. He called it a "generally accepted but neglected finding" that "performance on any recognition test can be manipulated by changing the characteristics of the distractors". Distractors are the wrong items in a recognition test. Thus the statement "recognition is easier than recall" is true only for a specified test with specified wrong answers.

When recall beats recognition

In 1973, Endel Tulving and Donald Thomson reported a result that the usual rule does not predict. Their paper states that "under certain conditions subjects consistently failed to recognize many recallable list words". They named this result recognition failure.

In their first experiment, 40 students at Yale University studied 24 pairs of words. Each pair had a cue word in small letters and a target word in capital letters, for example glue CHAIR. The cue was only weakly related to the target. The instruction was to remember the target words. Three of the later tasks are important here.

1. 1.The students got new words that are strongly related to the targets, for example table. They wrote the words that these new words made them think of. On average, they wrote 17.7 of the 24 target words.
2. 2.The students looked at the words that they had written. They circled each word that they recognised from the list. The mean was 4.2 words, "a hit rate of 24%".
3. 3.The final 10 of the 40 students got the original cue words, for example glue. They wrote the target words that they could recall. The mean was 15.2 of the 24 words, "a hit rate of 63%".

Those 10 students "recalled but did not recognize 69 words". Each of those words was in front of them in task 2, in their own writing, and they did not identify it.

Tulving and Thomson explained the result with the encoding specificity principle. McDermott and Roediger summarise the principle: a cue is effective to the extent that it "matches or overlaps the memory trace of an experience". The students had stored CHAIR together with glue. The word chair after the word table did not match that memory well. The cue glue matched it.

The condition is important. The experiment shows that recognition can fail. It does not show that recall is usually better. McDermott and Roediger give a second example, from a study of names. People sometimes do not recognise Shaw as the surname of a famous author. But the cue George Bernard frequently leads them to recall Shaw.

Recognition has two parts: familiarity and recollection

In 1980, George Mandler separated two types of recognition. The first type uses familiarity, which he called "context free". The second type uses recollection, which he described as "identification as a result of a retrieval process". These quotations are from a 2020 research note by Colin MacLeod on the history of Mandler's paper.

Mandler gave an example. You see a man on a bus. You are sure that you know him, but you do not know from where. You examine possible contexts: your work, a film, the television. The search can end with the answer "that's the butcher from the supermarket!" The first feeling is familiarity. The last step is recollection.

The example is now known as the butcher on the bus. Mandler's own words were "butcher-in-the-bus". MacLeod found a very similar example in a 1953 textbook by Charles Osgood, 27 years before Mandler's paper. Osgood's example is a man in a theatre queue who is "the man at the meat market". MacLeod suggests that the credit "should actually go to Osgood". He also asks a question about Mandler. Possibly the example was familiar to Mandler, and he did not recollect its source.

This section has a consequence for terminology. In memory research, recollection is not a second word for recall. Recollection is one of the two bases of recognition.

Are recall and recognition two different memories?

One study of people with amnesia is evidence against that idea. In 1992, Haist, Shimamura and Squire compared the two tests in amnesic patients. In amnesia, there is damage to a brain system that is important for conscious memory. The authors reported that "recall and recognition were found to be proportionately impaired in amnesic patients".

They concluded that the results "favor the view that recall and recognition are related functions of declarative memory". Declarative memory is conscious memory. In this view, recall and recognition are two tests of the same type of memory, and the tests supply different cues. This is one study, and it does not close the question.

What this means when you study

The first lesson is that a feeling of familiarity is not proof of knowledge. In a 2006 study, Roediger and Karpicke compared two groups of students. One group studied prose passages again. The other group took free recall tests on the passages. After 5 minutes, repeated study gave the better score. After 2 days and after 1 week, "prior testing produced substantially greater retention than studying, even though repeated studying increased students' confidence". When you read a definition again, the definition becomes familiar. That does not show that you can produce it.

The second lesson is that multiple-choice practice has a value. A common criticism is that multiple-choice tests "engage recognition processes rather than the productive retrieval processes known to enhance later recall". Jeri Little and her colleagues tested this criticism in 2012. They found that "properly constructed multiple-choice tests can indeed trigger productive retrieval processes". The condition was that the wrong answers were plausible. Plausible wrong answers let the students retrieve why each option was correct or incorrect.

For these two reasons, use the two types of test together. Our guide on how to memorise psychology terminology describes practice testing in more detail. Our article on spaced repetition explains how to put time between the sessions.

Psychology World has the two types of test. You save terms to a list, and then you drill that list. The Question game is a multiple-choice test, and thus a recognition test. It needs a list of at least four words. The Typing game is a recall test, because you type the answer yourself. A practical sequence is the Question game first and then the Typing game for the same list. You can see the study modes on the home page.