Luchins' Einstellung (Set) Experiments
Abraham S. Luchins (1942) was interested in the following: "Several problems, all solvable by one somewhat complex procedure, are presented in succession. If afterward a similar task is given which can be solved by a more direct and simpler method, will the individual be blinded to this more direct possibility?" Participants may develop a particular approach to a class ofsimilar problems. After the problems are changed so that this set (as in mindset) is no longer the most efficient method of solution, will people hang on to the set, or will they recognize the more direct method? The German word for set is Einstellung; our everyday problem-solving frequently involves an Einstellung:
We try our habitual ways of attacking a particular problem even though we may have more efficient procedures readily available to us.
Luchins usually used the water-jar problem in his experiments. People were given two or three jars of water of varying capacity and were supposed to figure out how to obtain a required amount of water by performing arithmetic operations on the volumes the jars would hold. For example, if jar A holds 6 gallons, jar B holds 15 gallons, and jar C holds 2 gallons, how can you obtain exactly 7 gallons? In this example, you could fill jar B (15), then fill jars A (6) and C (2) from the contents of B, leaving 7 gallons in B. The basic 11 problems appear in Table 13.1. To become familiar with the Einstellung effect, solve all 11 problems in order before you continue reading. In Luchins' first study, the first problem served as an illustration of the task. The appropriate solution is to take the larger jar (29) and subtract the smaller jar (3) three times to get the desired amount (20). Participants next solved problems 2 through 6, which may be considered set-establishing problems, because they all are most easily solved by the same method of solution.
In each case, the solution is to take the largest jar (always the middle one), subtract the first jar once, and finally subtract the last jar twice. If we label the jars A, B, and C from left to right, then the set the subjects developed for problems 2 through 6 can be represented as B _ A _ 2C. TABLE 13.1 Water-Jar Problems Luchins Used in His Einstellung Experiments Given the following empty jars as measures
Problem A B C
Obtain the required
amount of water
1 29 3 20
2 21 127 3 100
3 14 163 25 99
4 18 43 10 5
5 9 42 6 21
6 20 59 4 31
7 23 49 3 20
8 15 39 3 18
9 28 76 3 25
10 18 48 4 22
11 14 36 8 6
Besides the experimental or Einstellung group, there were two others.
One group, a control of sorts, began with problems 7 and 8 so that the experimenter could see how the group would solve them with no induced set solution. Another group was treated in the same manner as the Einstellung group, except that before problem 7, each participant wrote "Don't be blind" on the response sheet. Luchins had the participants write this phrase so that they would be cautious and not be foolish when they solved subsequent problems. Thus, the primary dependent variable was the number of participants using the Einstellung solution (B _ A _ 2C) on problems 7 and 8, even though problems 7 and 8 had much more efficient and direct solutions (A _ C for problem 7 and A þ C for problem 8). The results were that none of the control participants used the inefficient solution in solving problems 7 and 8; 81 percent of the Einstellung participants used the inefficient solution on 7 and 8; and the "Don't be blind" warning reduced the use of the Einstellung solution to about 55 percent. Furthermore, after doing problem 9, which cannot be solved via the set-inducing method, 63 percent of the Einstellung participants continued with the old solution on problems 10 and 11, whereas only 30 percent of the "Don't be blind" participants reverted to the set solution.
How reliable are the results from this experiment? Luchins' work was published before the use of statistical tests was common in psychological research, so statistical tests to establish reliability were not performed. However, such tests were largely unnecessary in this case, because Luchins provided us with evidence that his results could be replicated in other experiments. Many experiments are included in the original report, and the results are in general agreement with the Einstellung results just outlined. Altogether, Luchins tested more than 9,000 participants in his original studies. As noted previously, replication is an exceptionally convincing way to demonstrate the reliability of a phenomenon-the Einstellung effect has experimental reliability.
There are three types of replications: (a) direct, (b) systematic, and (c) conceptual. Direct replication is simply repeating an experiment as closely as possible with as few changes as possible in the method. Luchins replicated his original experiment several times with only slight changes in the participant population tested; such experiments constitute cases of direct replication.
In a systematic replication, all sorts of factors are changed that the investigator considers irrelevant to the phenomenon of interest. If the phenomenon is not illusory, it will survive these changes. So, for example, in a systematic replication of Luchins' experiments, we might vary the nature of the problems so that the set involves a different rule (or several different rules), the instructions are varied, the type of participants is varied, and so on. The Einstellung effect should be robust across all of these manipulations. If it is not, then we have found that variables previously thought to be irrelevant are actually important. This is crucial knowledge. In a conceptual replication, we attempt to replicate a phenomenon or concept in an entirely different way. In the language of Chapter 1, we simply 308 PART 4 • The Practice of Scientific Psychology use a different operational definition of the concept. Luchins examined other tasks besides the water-jar problem to establish the Einstellung phenomenon in diverse situations. He used series of geometry problems, words hidden in letters, and paper mazes.
Although the operational definitions of a problem and of a solution changed from experiment to experiment, in each case participants solved several problems that had a unique solution before they came to the critical problems that could be solved either by the Einstellung solution
or by a much more simple and direct solution. Just as in the water-jar experiments, participants usually used the old, circuitous solution and ignored the more efficient solution. These experiments constitute conceptual replications of the concept of Einstellung. Although Luchins' classic experiment is more than a half-century old, it is still being replicated in modern times (Schooler, Ohlsson, & Brooks, 1993) and cited in current literature (Bilalic_, McLeod, &Gobet, 2008).
You should note the following carefully: The problem of reliability of results is interwoven with the problem of generalizability and validity (see
the Application section). In general, researchers need to be careful about assuming that results will generalize to other subject populations. Most psychology experiments are conducted using white, middle-class college students as the subjects-hardly a representative group! College students are often the subjects of choice due to the ease with which they can be recruited for experiments (perhaps you are required to participate in experiments as part of your course requirements). While much research using college students does indeed replicate, we should be careful about assuming replication.
For example, psychologists have begun to study how non-Western cultures are different from those traditionally studied in psychology. Not only does self-concept vary in Eastern cultures (for example, Markus & Kitayama, 1991), so does the way people typically approach and solve problems (for example, Nisbett, Peng, Choi, &Norenzayan, 2001). Differences occur not only across cultures but also across ages. For example, older adults' performance often varies depending on the type of material used in the study. One study of off-topic speech found that older adults used more off-topic speech than younger adults but only when the older adults were discussing their own lives and not when they were describing pictures (James, Burke, Austin, &Hulme, 1998). Similarly, results on older adult conformity depend on whether the stimuli are emotionally meaningful (Pasupathi, 1999).
Psychologists are often eager to design interventions aimed at reducing the effects of aging; before doing so, however, they should be careful that their failure to obtain similar results with older adults is not simply due to the materials used in the experiment. As we progress from direct to systematic to conceptual replication, we show not only reliability but also increasing validity. Are we studying something that ties in with or is related to our knowledge of other psychological phenomena in a reasonable way? A conceptual replication is closely related to converging operations, which are procedures that validate a hypothetical construct used to explain behavior by eliminating alternative explanations
Question 1. Reread the description of Luchins' experiment outlined. There seem to be several problems with that study; see whether you can pinpoint two or three of them. As a hint, consider the following questions: Did Luchins use an appropriate control group? If not, what should be the control condition? Are there points in the study that require counterbalancing? What additional dependent variables might have been informative?
(response must be 300 minimum)
Questions 2: As a final, capstone question for this class, please consider all of the issues with which we wrestle in our society. Choose one and describe how we could address and answer this issue to some degree through the use of psychological research. (please not this question is not in any reference to the question listed above, it is a whole separate statement and the two do not correspond with one another.)
(response must be 300 minimum)