Processing Fluency

​

What is Processing Fluency?

Processing Fluency is the subjective experience of ease or difficulty with which a person processes information, used as a cue about the thing itself when that feeling is not set aside as coming from the print, the spelling, or the size of the request.

Fluency is defined as a subjective experience of ease or difficulty associated with a mental process. In other words, fluency isn’t the process itself but, rather, information about how efficient or easy that process feels.

That sentence is Box 1 of Oppenheimer’s 2008 review. The review’s opening sentence uses the same cut: fluency is the subjective experience of ease or difficulty with which we are able to process information. Song and Schwarz use the longer name, processing fluency, for that feeling when a font or a pronunciation produces it. This page keeps Processing Fluency for the feeling. It does not rotate the shorter word after this section. The 2008 font studies held the instructions constant. Students recalled details about as well in the hard font as in the easy font. They still gave the hard font a longer time estimate, a lower willingness rating, and, for the exercise, a lower fluency rating of the routine itself. The feeling of reading was available as a cue about the doing. Illusory Truth Effect is a higher truth rating after a plausible statement is heard again. A second hearing is not a font. Mere Exposure Effect is a more positive rating after a stimulus is shown again. One pass through a recipe is not that repetition.

​

Processing Fluency in 3 Depths

• Beginner: In the recipe study, 27 students read the same Japanese roll. Arial, 12 point, produced a mean estimate of 22.71 minutes, standard deviation 13.76. Mistral, 12 point, produced 36.15 minutes, standard deviation 15.3. The difficult-font mean carries a star, p ≤ .05. Cohen’s d, as printed, is 0.92. The authors call the gap 59 percent more time.
• Practitioner: Before treating a hard face as a harder task, read the memory row. The recipe memory means were 1.93 and 1.77 out of 2, d = 0.44, with no star. The exercise memory means were 1.3 and 1.2, d = 0.18, with no star. The estimate moved. The two-item check did not.
• Advanced: The exercise’s open minute line is 8.23 against 15.1, d = 0.9, and the 15.1 carries a dagger, p < .10. Willingness on that same page is 4.5 against 2.9 and is starred. A dagger is not a second copy of a star. The authors still write that the exercise was assumed to take 83 percent more time.

​

Origin

The feeling, and the shorter word fluency, are the 2008 review. Processing Fluency on this page is that feeling, named with the longer phrase Song and Schwarz print in the titles of the font paper and the pronunciation paper. Daniel M. Oppenheimer was at Princeton University, Department of Psychology, Green Hall, Princeton, NJ 08540. The review is Trends in Cognitive Sciences, volume 12, number 6, 2008, pages 237–241. The author PDF prints the DOI 10.1016/j.tics.2008.02.014. Box 1 says fluency is not the process itself. Box 2 says people try to attribute the feeling to a source, and that an obvious alternative source can lead them to set the feeling aside. Those boxes do not print a sample size. This page does not add a result from the papers those boxes cite. Song and Schwarz cite Alter and Oppenheimer, “Uniting the tribes of fluency,” as an unpublished manuscript, Princeton University, 2008. That manuscript was not re-read. No sentence from it is on this page. The font counts are Hyunjin Song and Norbert Schwarz, “If It’s Hard to Read, It’s Hard to Do: Processing Fluency Affects Effort Prediction and Motivation,” Psychological Science 19(10), pages 986–988. The line at the end prints received 4/10/08 and revision accepted 5/3/08. Read as American dates, those are 10 April 2008 and 3 May 2008. Both authors were at the University of Michigan, Department of Psychology, 530 Church Street, Ann Arbor, MI 48109-1043. The pronunciation counts are the same authors, “If It’s Difficult to Pronounce, It Must Be Risky: Fluency, Familiarity, and Risk Perception,” Psychological Science 20(2), pages 135–138. The end line prints received 11/5/07 and revision accepted 7/29/08, read here as 5 November 2007 and 29 July 2008. The recall counts are Schwarz, Bless, Strack, Klumpp, Rittenauer-Schatka, and Simons, “Ease of Retrieval as Information: Another Look at the Availability Heuristic,” Journal of Personality and Social Psychology 61(2), 1991, pages 195–202. The PDF read here does not print a DOI. Schwarz was at ZUMA in Mannheim. Bless was at the University of Mannheim. Strack was at the Max Planck Institute for Psychological Research in Munich. Klumpp, Rittenauer-Schatka, and Simons were at the University of Heidelberg. The note says the studies were based in part on diploma theses at Heidelberg, Experiment 1 by Simons and Experiment 2 by Rittenauer-Schatka, directed by Schwarz, Bless, and Strack. Grants Schw 278/2 and Str 264/2 came from the Deutsche Forschungsgemeinschaft. The paper quotes Tversky and Kahneman, 1973, page 208: people estimate a frequency by the ease with which instances come to mind. Availability Heuristic is that use of an easy example as a stand-in for a common event. The 1991 experiments ask what happens when the longer list is the harder list.

​

Key Points

Four cuts keep a minute estimate, a self-rating, and a hazard rating from inheriting one another. The font, the length of the list, and the pronunciation each have their own test.

​

Applications

Use the contrast when a hard font, a long list, or a hard name is treated as a property of the task, the person, or the substance. Keep each mean on the question that produced it.

​

Case Study

In the 2008 font paper, Song and Schwarz asked whether people misread the difficulty of processing instructions as the difficulty of doing the behavior. The behavior was not performed. Students at the University of Michigan read instructions and estimated the effort. Study 2 is the starred time contrast. Twenty-seven students read a recipe for a Japanese roll. The words did not change. The font did: Arial or Mistral, both 12 point. They estimated preparation time in open minutes and rated willingness to cook the dish. Two yes-or-no questions asked whether carrots and avocado were used. They then rated how easy the font was to read. The Arial mean was 22.71 minutes, standard deviation 13.76. The Mistral mean was 36.15 minutes, standard deviation 15.3. The table marks the difficult-font mean with a star, p ≤ .05, and prints d = 0.92. Willingness was 4.21 against 2.85, also starred, d = 0.79. Memory was 1.93 against 1.77, d = 0.44, with no star. Ease of reading was 6.14 against 2.69, d = 3.27, three stars. The authors’ prose says the same behavior was assumed to take 59 percent more time in the difficult font. That percentage is their summary of 36.15 against 22.71. It is not a second sample. Study 1, printed above Study 2 in the same table, is the exercise, and it is the reason the minute result does not travel as one number. Twenty students read identical exercise instructions in Arial or Brush, both 12 point. Willingness to adopt the routine was 4.5 against 2.9, starred, d = 0.95. The fluency index of the routine was 4.83 against 3.2, two stars. The open estimate of minutes was 8.23, standard deviation 5.61, against 15.1, standard deviation 9.28, d = 0.9. The 15.1 carries a dagger. The footnote prints dagger as p < .10, star as p ≤ .05, two stars as p ≤ .01, and three stars as p ≤ .001. The prose still says the exercise would take less time in the easy font, and that the same behavior was assumed to take 83 percent more time in the difficult font. The 83 percent sits next to the dagger. It is not a restatement of the recipe’s star. Study 3 used the recipe again and asked how much skill a restaurant cook would need. The means were 4.11 and 5.2, starred, d = 0.98. Those are not minutes, and the 19 students are not the 27. The lesson is a font, with the words held still, moving an estimate and a willingness rating while a two-item memory check does not show a starred drop. It is not a measured cooking time, and the exercise’s 15.1 minutes are not the recipe’s 36.15.

​

Boundaries and Failure Modes

Processing Fluency is the wrong summary when a dagger is treated as a star, when a longer list is treated as a stronger rating, or when a hard name is treated as a tested ingredient. The first boundary is the mark on the minute line. Study 2’s 36.15 is starred. Study 1’s 15.1 is a dagger, p < .10. The authors discuss both as time results and attach 83 percent to the exercise and 59 percent to the recipe. A reader who exports one “hard font adds about 60 to 80 percent” has merged a p ≤ .05 row with a p < .10 row, and has merged an exercise with a recipe. Study 3’s skill rating does not belong in that merge. The standard deviations are large beside the means: 5.61 and 9.28 on the exercise minutes, 13.76 and 15.3 on the recipe minutes. Those spreads are the samples. They are not a timer. The second boundary is the recall experiments, and it is where a message about other people fails to do the work the design hoped. Experiment 1’s interaction is p < .07. Experiment 2’s collapsed interaction is p < .02, and the paper says the pattern replicated. The diagnosticity message in Experiment 2, telling students that most earlier participants had found the task easy or hard, did not yield a significant triple interaction. The paper prints F below 1 for that triple test. A separate simple interaction was significant under high diagnosticity, F(1, 76) = 5.98, p < .03, and not under low diagnosticity, F below 1. The paper reports both facts. Table 2 prints the split: under low diagnosticity the cells are 5.1, 5.0, 4.7, and 5.2; under high diagnosticity they are 5.2, 4.5, 4.7, and 5.4, for six assertive, six unassertive, twelve assertive, and twelve unassertive, in that order. The prose means 5.2, 4.7, 4.7, and 5.3 are the collapse. They are not a fifth sample. Judges scored the last two examples of five students per condition. Assertive examples averaged 9.6 and unassertive examples 2.8 on a 1-to-11 scale, F(1, 32) = 663.0, p < .001. The last two of twelve were not milder copies of the last two of six. The twelve-example unassertive pair averaged 2.3 against 3.3, F(1, 32) = 4.26, p < .05. Diluted content does not explain the drop. The check used five students per condition, not the whole college sample. The third boundary is the common misuse, including the hope that naming the source always turns the feeling off, and the hope that a hard name is only “more dangerous.” Experiment 3 played meditation music over headphones and told students the music could help or hinder recall. The triple interaction is F(1, 70) = 9.75, p < .001. Under high diagnosticity the simple interaction is F(1, 70) = 5.97, p < .02. The within-valence tests are one-tailed: assertive examples, 5.6 against 4.5, t(70) = 1.55, p < .06; unassertive examples, 3.1 against 4.4, t(70) = 1.91, p < .03. The paper says this pattern did not produce a crossover. Under low diagnosticity the simple interaction reverses direction, F(1, 70) = 3.88, p < .06. The table prints 3.9 and 5.1 for six examples, assertive then unassertive, and 4.8 and 4.0 for twelve. The prose prints 5.11 where the table prints 5.1 for the low-diagnosticity unassertive cell of six. This page does not average 5.1 and 5.11. The ease check in Experiment 3 was not significant: six examples 5.9, twelve examples 6.7, F(1, 70) = 1.76. The scale is the mean of two 9-point questions, and n is 9 or 10 per condition. Those means are not Experiment 1’s 6.3. In the pronunciation paper, Study 3’s 35 students rated amusement-park rides. Hard names were riskier when the risk was getting sick, 4.35 against 3.02, t(18) = 4.36, p < .001, d = 1.48, and more adventurous when the risk was excitement, 4.04 against 3.06, t(15) = 2.94, p = .01, d = 1.08. The interaction with desirability was F(1, 33) = 0.61, p = .44. A hard name did not mean “danger” alone. None of these figures is a kitchen, a gym, a self-review at work, or a labeled ingredient.

​

Common Misconceptions

Three retellings turn a dagger into a workout, a longer list into a stronger trait, or any easy sentence into a second hearing.

​

Related Concepts

The font estimate sits next to a truth rating after repetition, a judgment that uses an easy example, liking after a repeated showing, a forecast of one’s own work, and a faster decision that two strings are words.

​

One-Line Takeaway