Experimental Psychology

The scientific investigation of basic behavioral processes including sensation, emotion, and motivation, as well as such cognitive processes as perception, memory, learning, problem-solving, and language.

Experimental psychologists work to understand the underlying causes of behavior by studying humans and animals. Animals are studied within and outside laboratory settings for a variety of reasons. A researcher may wish to learn more about a particular species, to study how different species are interrelated, to investigate the evolutionary significance of certain behaviors, or to learn more about human behavior.

Experimental psychology flourished in the second half of the nineteenth century with the work of such figures as G. T. Fechner (1801-1887), whose Elements of Psychology (1860) is considered the first study in the field, and Wilhelm Wundt (1832-1920), who established the first psychological laboratory in 1879. Others, including Hermann Ebbinghaus and E.B. Titchener (1867-1927), used laboratory methods to investigate such areas as sensation, memory, reaction time, and rudimentary levels of learning. While controlled laboratory studies continue to make major contributions to the field of psychology, experimental methods have also been used in such diverse areas as child development, clinical diagnosis, and social problems. Thus, the concept of experimentation can no longer be limited to the laboratory, and "experimental psychology" is now defined by method and by the kinds of processes being investigated, rather than its setting.

An experiment in any setting tests a hypothesis, a tentative explanation for an observed phenomenon or a prediction about the outcome of a specific event based on theoretical assumptions. All experiments consist of an independent variable, which is manipulated by the researcher, and a dependent variable, whose outcome will be linked to the independent variable. For example, in an experiment to test the sleep-inducing properties of the hormone melatonin, the administration of the hormone would be the independent variable, and the resulting amount of sleep would be the dependent variable.

In simplest terms, the effects of the independent variable are determined by comparing two groups which are as similar to each other as possible, with the exception that only one group has been exposed to the independent variable being tested. That group is called the experimental group; the other group, which provides a baseline for measurement, is called the control group.

Although ideally the experimental and control groups will be as similar as possible, in practice, most psychological research is complicated by a variety of factors. For example, some random variables—differences in both the subjects themselves and in the testing conditions—are unavoidable and have the potential to disrupt the experiment. In addition, many experiments include more than one group of subjects, and establishing a true control group is not possible. One method of offsetting these problems is to randomly assign subjects to each group, thus distributing the effect of uncontrollable variables as evenly as possible.

The subjects' attitudes toward the experimental situation is another condition that may influence the results. This phenomenon is best demonstrated by what is referred to as the placebo effect. Subjects in experiments that test medical and psychological treatments often show improvement solely because they believe the treatment has been administered. Thus, the administration of a placebo (a supposed treatment that in fact contains no active ingredient) to a control group can disclose to the experimenter whether improvement in the subjects' conditions has been caused by the treatment itself or only by the subjects' belief that their condition will improve. Interference may come from an additional variable, experimenter bias, the unintentional effect of the experimenter's attitudes, behavior, or personal interests on the results of an experiment. The experimenter may, for example, read instructions to two groups of research subjects differently, or unintentionally allow one group slightly more or less time to complete an experiment. A particularly powerful type of experimenter bias is the self-fulfilling prophecy, whereas the researcher's expectations influence the results. In a well-known example, when laboratory assistants working with two groups of randomly selected rats were told that one group was brighter than the other, they treated the rats in such a way that the supposedly "brighter" group learned to negotiate a maze faster than the other group. Subtle differences in the assistants' handling of the "brighter" group had produced the results they were conditioned to expect.

In experiments utilizing a placebo, experimenter bias may be prevented by a double-blind design, in which not only the subjects but also the persons administering the experiment are unaware of which is the control group and what results are expected. In general, experimenters can minimize bias by making a vigilant attempt to recognize it when it appears, as well as resisting the temptation to intentionally influence the outcome of any experiment. The results of experiments are generally presented in a report or article that follows a standard format of introduction, method, results, and conclusion.

Experimental research can also be conducted through quasi-experiments, studies which lack the control of a true experiment because one or more of its requirements cannot be met, such as the deliberate use of an independent variable or the random assignment of subjects to different groups. Studies of the effects of drugs on pregnant women, for instance, are based on data about women who have already been pregnant and either taken or not taken drugs. Thus, the researcher has no control over the assignment of subjects or the choices with which they are presented, but he or she can still measure differences between the two populations and obtain significant findings. These findings gain validity when they are based on data obtained from large numbers of subjects and when their results can be replicated a number of times. Such studies provide a basis for investigations that would otherwise be impossible.