> ## Documentation Index
> Fetch the complete documentation index at: https://meta.niceshare.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Von Restorff Effect

> Von Restorff Effect is better memory for the one different item among similar items. Her early-item average was .70 against .40 on an unrelated list.

<Info>
  **Category**: Effects<br />
  **Type**: Better memory for the one item that differs in kind from similar neighbors than for that item among its own kind<br />
  **Origin**: Hedwig von Restorff, *Psychologische Forschung* 18 (1933), pages 299–342, as described by R. Reed Hunt, *Psychonomic Bulletin & Review* 2(1), March 1995, pages 105–112. The 1933 paper was not re-read here<br />
  **Also known as**: isolation effect
</Info>

<Note>
  **Quick Answer** — Von Restorff Effect is better memory for the one item that differs in kind from the similar items around it than for the same item surrounded by its own kind. Hedwig von Restorff reported that pattern in 1933. The average Hunt prints for her early-position test is a second comparison: .70 on the isolation list and .40 when that item sat in an unrelated list. A red background on a letter, in a 1972 test, was a smaller lift than changing the letter into a number.
</Note>

## What is Von Restorff Effect?

Von Restorff Effect is better memory for the one item that differs in kind from the similar items around it than for the same item surrounded by its own kind.

> The isolated item was not perceived as unusual and was not particularly salient to the subject.

That sentence is Hunt's English of von Restorff, 1933, page 319. The syllable or the number sat in the second or third place in a list of ten. At that point, she writes, the person could not yet know the rest of the list. The test was built so the odd item would not pop out as a surprise in the middle. [Primacy Effect](/effects/primacy-effect) on this site is extra weight on the first item because it is first. Her comparison keeps the serial place matched and changes the neighbors.

A tray of nine white cups and one blue cup is the shape of the similar-neighbor list. A tray on which every object is a different kind is the unrelated list. Remembering the blue cup is the first arrangement. It is a different question from remembering that cup when nothing else matches. Her materials were nonsense syllables, numbers, and the other kinds named below. The cups are only the shape.

### Von Restorff Effect in 3 Depths

* **Beginner**: One number among syllables is easier to recall than that number would be among other numbers. In the early-position test Hunt describes, the odd item was near the start. It was not printed in red.
* **Practitioner**: Name the neighbors and the serial place before calling an item isolated. In a 1972 letter list, position 5 as a number was 28.5 points above the control letter. The same position as a white letter on a red field was 6.9 points above that control, under serial scoring.
* **Advanced**: The .70 and the .40 are her average for an early isolate against an unrelated list, after 10 minutes of reading, in 15 people. Hunt's repeat split the isolate: a syllable at .70 and a digit at .80, against .40 and .30. He prints no test of those four proportions. The red field is a third operation.

## Origin

The paper is Hedwig von Restorff, "Über die Wirkung von Bereichsbildungen im Spurenfeld," *Psychologische Forschung* 18 (1933), pages 299–342. This page did not open that German text. The account below is R. Reed Hunt, "The subtlety of distinctiveness: What von Restorff really did," *Psychonomic Bulletin & Review* 2(1), March 1995, pages 105–112. The PDF read here does not print a DOI. The publisher record is 10.3758/BF03214414. The manuscript was received on 11 April 1994 and accepted on 17 November 1994. Hunt was at the University of North Carolina at Greensboro. A footnote says the work was supported by NICHHD grant HD256587, and that Andrea Dorsch translated the 1933 paper. This page did not open that translation. Hunt says the 1933 paper had not been published in English.

Hunt writes that von Restorff worked as a postdoctoral assistant to Wolfgang Köhler at the Psychological Institute of the University of Berlin, up to the time Köhler resigned in protest against Nazi interference with the Institute. Citing Mary Henle (1986), he writes that the incident which precipitated the 1935 resignation was the dismissal of the postdoctoral assistants, including von Restorff and Karl Duncker. Henle's book was not re-read here. No page from it is added.

During that Berlin period she published two papers. The first is the 1933 dissertation. Hunt says the second, co-authored with Köhler, is about spontaneous reminding and the role of intentionality in the memory test. His reference list dates that paper 1935 and prints *Psychologische Forschung* volume 19, pages 56–112. The publisher record for Köhler and von Restorff, "Analyse von Vorgängen im Spurenfeld," prints volume 21, pages 56–112, issue date December 1937, received 26 November 1935. This page did not re-read that paper. The two records are left side by side. No cell from the 1937 text is used here.

Hunt also writes that she neither pioneered the isolation paradigm nor treated vividness as the explanation. He names Mary W. Calkins (1894, 1896), Arthur Jersild (1929), and W. L. van Buskirk (1932) as earlier English-language isolation studies aimed at vividness. Those papers were not re-read. No percent from them is on this page. Her own interest, as he recounts it, was interference. Similar massed items agglutinate in memory. The item that does not share that similarity stays distinct. She then asked whether that advantage required the item to be perceptually salient at the moment it appeared.

## Key Points

Four cuts keep a highlighted line from inheriting her average. The neighbors, the early place, the unrelated list, and the method of the change each have their own number.

<Steps>
  <Step title="Compare the odd item with its own kind">
    The effect associated with her name, in Hunt's opening account, is better memory for a syllable embedded in numbers than for the same syllable in the same serial place of a homogeneous list. An early paired-associate test used eight pairs. Four pairs were one material. Each of the other four pairs was the only pair of its material. She found better recall for the isolated pairs. Hunt says she does not report recall of the isolated pair by serial position. This page does not invent that split.
  </Step>

  <Step title="Put the odd item where the list is not yet known">
    In the test whose average Hunt prints, each list had ten items. The isolated item was a number among nine nonsense syllables, or a nonsense syllable among nine numbers. It stood in the second or the third position. She does not say how often it was second rather than third. People were instructed to remember. Each item was shown for 1.5 seconds. They then read a prose passage for 10 minutes, recalled the list, and recalled the passage. Fifteen people each saw three lists, on three successive days. The unrelated list came on day 1, so the later isolate would be camouflaged.
  </Step>

  <Step title="Keep the unrelated list as its own bar">
    Her argument, as Hunt translates page 313, is that a syllable among nine numbers is very different from those numbers, and a syllable in a list of ten unrelated items is still that different from its neighbors. If the syllable is remembered better in the first list, similarity among the other items is doing work that raw difference does not. She reported average correct recall of .70 for the isolation list and .40 for the control list. Hunt's between-subjects repeat, with the isolate always in position 2, found .70 for an isolated syllable and .80 for an isolated digit, against .40 and .30 on the unrelated lists. [Clustering Illusion](/effects/clustering-illusion) is a streak in a small chance sample read as a cause. One remembered syllable is not that streak.
  </Step>

  <Step title="Do not price a red field as a change of kind">
    Neal E. A. Kroll, at the University of California, Davis, isolated position 5 in lists of eight letters in four ways: an oval surround, a red background, a larger letter, or a number. Under serial scoring, the red background and the surround were each 6.9 points above the control letter. The number was 28.5 points above it. [Priming Effect](/effects/priming-effect) is a faster correct decision that two letter strings are words when the words are common associates. A millisecond gap between word pairs is not a point gap in a letter list.
  </Step>
</Steps>

## Applications

Use the contrast when someone treats a highlighted line as a measured gain in memory. Keep .70, .40, and the 1972 point gaps inside the lists that produced them.

<CardGroup cols={2}>
  <Card title="Name the neighbors before the highlight" icon="list-ol">
    A line is isolated when the other lines share a kind and this one does not. Write that kind down. In her early-position list the kind was nonsense syllables or numbers, nine of them, with one item of the other kind. A slide that marks one bullet in red has not yet said whether the other bullets are the same kind of thing.
  </Card>

  <Card title="Match the serial place" icon="arrow-down-1-9">
    Her printed average uses position 2 or 3 of ten, and she did not split those two places. Hunt's repeat used position 2 only. Kroll used position 5 of eight, after four letters had already set the list. A team that moves the odd price to the front, or buries it in the middle, has changed the test. Ask which place was actually shown.
  </Card>

  <Card title="Separate a color from a category" icon="palette">
    Kroll's red condition was a white letter on a vivid red field, not red ink. The enlarged letter was 72 mm high against letters of 20 mm. The number came from 2 through 9. The control letter at position 5 was recalled on 45.6 percent of serial-scored trials. Budget a color change with the 6.9-point row. Budget a change of kind with the 28.5-point row. Do not swap the rows.
  </Card>

  <Card title="Score the rest of the list" icon="clipboard-check">
    Kroll's serial average for positions 1–4 and 6–8 was 56.8 percent on control lists. The number lists sat 0.8 points from that average. The red lists sat 2.7 points from it. The large gap is the isolated slot. Position 6, the next item, was slightly higher under serial scoring, 53.7 percent against a control of 48.8, not lower. A highlight that is sold as "and the neighbors disappear" is not this table.
  </Card>
</CardGroup>

## Case Study

Hunt repeated the early-position comparison at the University of North Carolina at Greensboro and set it beside the average he attributes to von Restorff. Her procedure, as he describes it, used 15 people and three lists on three successive days. Everyone saw the unrelated list on the first day and an isolation list on each of the next two days. The unrelated list was shown first so the later isolate would not be perceptually salient. Each list had ten items, shown one at a time for 1.5 seconds, under intentional memory instructions. The unrelated items were a number, a syllable, a color patch, a single letter, a word, a photograph, a symbol, an actual button, a punctuation mark, and the name of a chemical compound. People recalled the name of each item: the color name, the word "button," or what the picture showed. The isolation lists were nine nonsense syllables and one number, or nine numbers and one nonsense syllable. The isolated item was in the second or the third position. After the list, 10 minutes went to reading a prose passage that would itself be tested. Free recall of the list came next, then recall of the passage.

The measure Hunt treats as decisive is correct recall of the critical item by list type. Von Restorff's average, in his report, was .70 on the isolation list and .40 on the control list. He does not print a separate syllable row and number row for that .70, and he does not print a statistical test. He also notes that she does not say how often the isolate was in the second position rather than the third. With the item third, two earlier items might already have supplied a context. He gives two reasons he does not rest the result on that worry. The unrelated list came first as camouflage. His own repeat put the isolate only in the second position, and the proportions stayed close to hers.

The repeat changed two things. Lists were between subjects, so a person saw only an isolation list or only an unrelated list. The isolate was always second. Forty undergraduate volunteers took part, 20 in each condition. Two isolation lists were built, one with nine nonsense syllables and one digit, the other with nine digits and one nonsense syllable. Each list was shown to half the people in the appropriate condition. The unrelated lists put a digit or a syllable in position 2. After one list and 10 minutes of reading prose, people freely recalled the list. Correct recall was .70 for an isolated syllable and .80 for an isolated digit. On the unrelated lists it was .40 for syllables and .30 for digits. Hunt calls this a close replication despite the procedural changes. He does not print a test of the four proportions, and this page does not add one.

The lesson is a recall gap for an early item of the other kind, against the same kind of item in an unrelated list, after a filled 10 minutes, in these two procedures. It is not a rate at which shoppers remember a red price, and it is not a rate at which students remember a highlighted sentence. Fifteen and 40 are different samples. The .70 on her isolation list is an average across the syllable lists and the number lists. The .70 in his repeat is the syllable cell alone.

## Boundaries and Failure Modes

Von Restorff Effect is the wrong summary when a red mark is treated as her .70, when an unrelated list is treated as if difference had been removed, or when a letter test at half a second is treated as a 10-minute memory for syllables.

The first boundary is salience. Hunt writes that she used a single mid-list isolate in only one of her experiments, a recognition test. That list had 20 items, 19 syllables and one number, or 19 numbers and one syllable. She moved to that sharper contrast because recognition tests with several isolated items had produced smaller differences than her recall data. Hunt says the sharper contrast then made the recognition gap larger than the recall gap. He prints no recognition percent. This page does not invent one. Her stated aim for the early-position recall test was to avoid the case in which the critical item would stand out as perceptually unique. Hunt cites Pillsbury and Raush (1943) for an isolation effect with the isolate in the first position. That paper was not re-read. No cell from it is on this page. A design that waits until the middle, then changes color, is the design later attention theories were written to explain. It is not the design behind the .70.

The second boundary is the size of the change, and Kroll measured it directly. *Psychonomic Science* 26(6), 1972, pages 333–334, prints the work. The PDF read here does not print a DOI. The publisher file is 10.3758/BF03328635. Thirty university students each saw 80 lists, 16 in each of five conditions, in one shared list order with the condition counterbalanced. Letters appeared at two per second. Recall was immediate and in order, with a guess required when an item was missing. People were warned that one item would sometimes differ in one of four listed ways. The position of that item was not announced. Serial scoring counted an item only in the correct position. Even the smallest lift, the oval surround, was above the control: t(29) = 2.38, p \< .05. An analysis of variance gave F(4, 116) = 28.47, p \< .01. Tukey's test at the .01 level had an honestly significant difference of 9.7 percent. The enlarged letter beat the red background by 9.8 points. The number beat the enlarged letter by 11.8 points. Free-recall scoring, which ignores position, put the surround at +8.5 and the red field at +4.0. He reports that interaction as t(29) = 2.10, p \< .05, and calls it small. Subjects said the enlarged letter and the surround were more attention-getting than the number. He prints no rating count. He also says people rarely wrote an incorrect number. He prints no count of those rare answers.

The third boundary is the export, and it is the common misuse. The .70, the .40, the .80, and the .30 belong to syllables and digits in the procedures above. The 6.9, the 16.7, the 28.5, and the control 45.6 belong to immediate serial recall of eight letters at the University of California, Davis. They are not a web button, a shelf, a classroom, or a meeting. [Availability Heuristic](/effects/availability-heuristic) is a rate judged by how easily an example comes to mind. Ease of thinking of one odd price is not Kroll's point gap. [Frequency Illusion](/effects/frequency-illusion) is the belief that a noticed thing happens a whole lot. Remembering the odd item is not a count of how often that item occurs outside the list.

## Common Misconceptions

Three retellings move the effect onto a color, onto any difference at all, or onto a loss for the next item. The 1933 average and the 1972 table answer different questions.

<AccordionGroup>
  <Accordion title="She found the effect by printing one word in red">
    The recall test behind the .70 placed a syllable or a number in position 2 or 3, after an unrelated list had already been seen, so the odd item would not be salient. A single item isolated in the middle appears in her recognition experiment, and Hunt prints no percent for it. Kroll's red condition, forty years later, was a white letter on a red field in position 5, and the serial lift was 6.9 points. The number in that same slot was 28.5 points. A red word is one later method. It is not her average, and it is not the largest row in the letter table.
  </Accordion>

  <Accordion title="Any item that differs is remembered at the isolation rate">
    Her page 313 argument, in Hunt's English, is that the syllable is just as different from unrelated neighbors as it is from a row of numbers. Recall was still higher in the number row: .70 against .40 in her average, and .70 or .80 against .40 or .30 in his split. Difference without a similar background was the lower bar. Kroll's four methods were all differences, and they were not one rate. The first trial of each condition used six people, and he notes that a single error moves that percent by 16.7. Those cells stay off this page.
  </Accordion>

  <Accordion title="The odd item is remembered and the next item is lost">
    Kroll's serial scoring found a small positive spread to position 6: the mean of the isolated lists was 53.7 percent against a control of 48.8, t(29) = 2.28, p \< .05. Free-recall scoring found no spread to that position. The average of the other positions barely moved. Her 1933 report, as Hunt gives it, does not print a neighbor-by-neighbor loss. A claim that isolation wipes the next line is not either table.
  </Accordion>
</AccordionGroup>

## Related Concepts

The recall gap sits next to extra weight on a first item, a judgment from an easy example, a faster word decision, a clump read as a cause, and a belief that a noticed thing happens a whole lot.

<CardGroup cols={3}>
  <Card title="Primacy Effect" icon="list-ol" href="/effects/primacy-effect">
    Extra weight on the first item because it comes first. Her isolate was in position 2 or 3, matched against the same place in another list. The gap is the neighbors, not the privilege of being first.
  </Card>

  <Card title="Availability Heuristic" icon="brain" href="/effects/availability-heuristic">
    A rate judged by how easily an example comes to mind. Remembering one odd item from a list of ten is a recall score for that item. It is not a rate in the world outside the list.
  </Card>

  <Card title="Priming Effect" icon="font" href="/effects/priming-effect">
    A faster correct decision that two letter strings are words, when the words are common associates. The 1971 gap is in milliseconds. It is not a point gap between a number and a letter.
  </Card>

  <Card title="Clustering Illusion" icon="chart-scatter" href="/effects/clustering-illusion">
    A streak or a pile in a small chance sample, read as proof the process is not chance. One remembered syllable among numbers is a single item's recall. It is not a grid of hits.
  </Card>

  <Card title="Frequency Illusion" icon="eye" href="/effects/frequency-illusion">
    The belief that a phenomenon happens a whole lot once it has been noticed. Recalling the odd item more often than the control item is a score inside the list. It does not show that the odd item has become common outside it.
  </Card>
</CardGroup>

## One-Line Takeaway

<Tip>
  Before you trust a highlight, write the neighbors' kind, the serial place, and the recall of that same item when the neighbors do not match.
</Tip>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.