Bizarreness Effect: Why Odd Sentences Outlast Ordinary Ones

The effect is about contrast rather than content. A sentence describing an elephant balancing a teapot is not easier to store than a sentence about a gardener and a bucket; it is easier to find later, because when the list is scanned at recall it does not look like anything else in it. That is the distinctiveness account, and it is the one McDaniel and Einstein argued for. The alternative explanations, such as bizarre items simply getting more attention or more study time, were tested in their experiments and did not hold up.

The condition that follows from distinctiveness is the part most summaries leave out. Distinctiveness is a relation, not a property. An item can only stand out from a background, so the effect appears in mixed lists, where bizarre and ordinary items compete for the same recall attempt, and not between separate groups studying one kind of list each with nothing else in between. The background does not have to be the study list: the fifth experiment in the same paper recovered the advantage between lists when the students learned common imagery afterward. What removes the effect is the absence of ordinary material to stand out from, not the particular design that supplies it.

Where in memory the advantage sits has been examined directly. Riefer and Rouder applied multinomial modeling to separate storage from retrieval in 1992 and placed the benefit in retrieval: the bizarre items are not stored better, they are easier to get back to. That fits the mixed-list condition, since retrieval is where the items of one list compete and storage is not.

Two limits are worth holding onto. The advantage is most reliable in free recall, where the person has to generate the items without help, and it is less consistent when the test supplies a cue or asks only for recognition, which is a sign that the gain is about access rather than strength. And the memory benefit attaches to the strange element itself, not to everything attached to it: the elephant and the teapot come back, while the point the sentence was supposed to carry may not. Reviews of the literature treat the effect as real, modest and heavily conditional rather than as a general-purpose memory trick.

Mark McDaniel and Gilles Einstein published "Bizarre imagery as an effective memory aid: The importance of distinctiveness" in the Journal of Experimental Psychology: Learning, Memory, and Cognition in 1986. It reports five experiments with 304 undergraduates in total. The materials were sentences built around three nouns, each written in a bizarre and a common form, so that the two versions carried the same words in different relations.

Experiment 1 set up the comparison the paper is known for. When the two kinds of sentence were mixed in one list, so that each student studied both, the bizarre sentences were recalled better. When the kind of imagery was varied between lists, so that one group saw only bizarre sentences and another only common ones, the advantage was not found. Nothing about the sentences had changed; what changed was whether a bizarre sentence had ordinary ones around it.

The remaining experiments ruled out the obvious alternatives. Experiment 2 found the effect under imagery instructions but not under semantic processing instructions. Experiments 3 and 4 varied how long each sentence was shown and whether students set their own pace, and the pattern of results did not follow the pattern an attention account predicts. Experiment 5 found a between-list advantage for bizarre imagery, but only when the additional learning involved common imagery, which again puts the ordinary material in the role of the backdrop and shows that it need not sit in the study list itself. The authors concluded that distinctiveness, rather than extra attention, underlies the effect.

The result has held up as a conditional one. Later work, including Riefer and Rouder's modeling in 1992 and Einstein, McDaniel and Lackey's study of interference in 1989, has refined where the advantage sits rather than overturned it, and reviews of the distinctiveness literature describe the bizarreness benefit as dependent on a setting that supplies ordinary material to contrast with, on free recall, and on the strange item being rare enough in that setting to count as odd.

The rows below are places where strangeness competes with the rest of the material, not findings from the studies cited here.

The effect is usually described as a memory aid, so the useful advice has two halves: how to get the benefit, and how to stop it distorting what you remember.

• Keep strangeness rare. One odd item in twenty is distinctive; twenty odd items are a list of odd items. If you are building a mnemonic, use the trick on the few things that keep escaping you.
• Attach the oddity to the thing you need, not beside it. What comes back is the strange element, so the fact, the name or the brand has to be part of it.
• Expect uneven recall rather than better recall. In a mixed list the odd items dominate the recall attempt, which is a redistribution of what comes back and not a bonus on the total.
• When you are the audience, notice that the vivid example is doing the remembering for you. Ask what the ordinary evidence was and go back for it.
• For material that must be recalled completely, such as a procedure or a checklist, drop the bizarre entirely and use order, grouping and repetition instead.
• In records that matter, write the routine details down. They are the ones a striking event pushes out.

Three entries all describe one item standing out from a list, and they differ in what does the standing out.

• It is not the Von Restorff effect, where an item is isolated by its form rather than its meaning: a word in red among black ones, a number among words. Both run on distinctiveness, but bizarreness is in what the sentence describes and isolation is in how the item looks.
• It is not the humor effect, the better recall of material that is funny. Funny material is often strange, which makes the two hard to separate, but humor adds the reward of the joke and the extra processing it takes to get it.
• It is not the picture superiority effect, the better memory for images than for their printed names. That comparison is about the format the material arrives in, not about how unusual its content is.