III. Issues Pertaining to Knowledge
This section (i.e., A.) is here for conceptual clarity. Many of the details of the specific items are the things covered in this course and therefore this section is most completely understood after having read the entire manuscript.
What makes something true. What makes something understood. What are considered facts. Only after facts are separated from illusion can the systematic collection of facts proceed. These issues were developed under the "Truth and Understanding" headings in the earlier portion of this chapter.
a. Most Global Context (field)
Chapter 2 presented detailed information on molarity and time scale. Once a person chooses a level of molarity and a time scale, "Psychology of Learning" for example, rarely are other chosen fields or other time scales considered.
i. Molarity, Level of Analysis, or Unit Domain
ii. Time Scale of Adaptation or Time Domain
b. Context (research specialty or area)
In point of fact, a researcher rarely steps outside a specific context or research specialty. Examples of these "local" contexts would be "matching" or "timing."
c. Immediate Context (researcher's own lab)
Sometimes research is carried out with very little connection to any other research at all. This can be OK, but it can also spell the doom of a lot of work. The work can be so idiosyncratic that virtually no one but the investigator (or his students) cares about or can benefit by the knowledge. When doing research, it may be a good exercise to consider how the research bears on each of the above contexts.
It may be that it interests us and that is a good enough reason. If it interests us, then that means some implicit theory we believe did not prepare us for the event that piqued our interest.
This type of knowledge gathering systematically obtains facts as well as the necessary other information to develop a coherent frame of reference or context for meaningful explanation. The following enumeration illustrates the kinds of procedures or kinds of information which could be obtained.
a. Relevant / Irrelevant Variables
What are functionally significant variables and which are irrelevant or are simply confounds?
b. Parameter Documentation
What changes in the dependent variable are caused by what changes in the independent variable?
What other well-understood functional relationships have properties which are similar so that even more fundamental explanations common to both phenomena can be uncovered?
If we deduce some experimental test such as: “according to my understanding of the processes involved, if we double the reinforcement rate then the rate of behavior should be halved. Keep in mind that a single positive finding supports a theory but only marginally. A single negative finding whose interpretation is correct is very damaging to a theory. But also keep in mind that everything hinges on the author's interpretation of the disconfirmation. Multiple converging evidence must be a requirement of theory testing.
The degree to which research is integrated within its paradigmatic context determines two important aspect of the research, its likely validity and its usefulness.
Research can be either more or less well integrated into its paradigmatic context.
a. Multiple Converging Support
Research well integrated within a paradigmatic context has a wide variety of support, while nonparadigmatic research has only itself to provide proof. Integration provides support for the validity of the position.
b. Predictable and Understood Generalizations
Research well integrated within a paradigm also has many aspects which are “pre-understood.” Its underlying machinery has generally already been thought through. Relationships are, therefore, understood. Integration enables the information to be used more effectively.
Research can produce a single fact or a large set of interrelated findings. This is not the degree of integration into the paradigm, but rather the degree of integration with "itself."
We may want to find out what happens if we do x to our subjects. This is a single fact.In this case, it is determined that level x of independent variable y will have z effect. 5 grams of food results in 50 pecks.
The task here is to do enough research to understand how something works across its whole range. In this case, it is determined that the whole family of levels which independent variable y can take can be described by equation 2. This is illustrated in the figure below. Note the difference between the information it provides and the information provided by an isolated treatment (the above figure).
This refers to the degree to which an event is taken at face value or is seen as an instance of a more fundamental process. Do you see a pigeon pecking on a red key, or do you see an operant maintained by its consequences? The knowledge sought can be simple "at face value" information, such as “John jumped when I said boo.” Or the knowledge sought can be a general rule, such as “sudden stimuli produce startle responses.” Or your interest may be even broader, such as “some stimuli cause unconditioned responses.” In order to generalize an event to a general class, you must have some paradigm within which to view the event. It is the paradigm that gives a finding its generality. Fleming could not have discovered penicillin if he did not clearly understand the contradiction of the event he observed and the theoretical significance of that contradiction.
In the absence of any abstraction, the actual behaving organism is what you are watching and you want to know, for example, if it will move to the right or to the left when you "poke" it. Precisely why that seems like an interesting thing or what it means is difficult or impossible to articulate. This type of question is very concrete. It tends to be interesting because of some implicit connection to some implicit paradigm.
At the other extreme of abstraction is considering some specific behavior as a representative of something else. A specific research question may be implemented with some specific subject and some specific procedure, but they are seen as arbitrary (other than the necessity that the model must accurately reflect the properties of interest in the target). Experimental paradigms, which are used to provide information on some other specific situation, or on all situations, are labeled "models."
There are a countless number of ways two things can be the same or different: same size, same weight, same proportions, same color, same material, same location, etc. A model must be the same as the target with respect to relevant features. It does not matter whether there are similarities with respect to the irrelevant features. In fact, a model is very often used to eliminate some undesired features. For example, we may make a model airplane because the size and cost of a real one make it prohibitive for children to play with.
Those features of a model
of specific interest are relevant. A model of a wing must have the same lift,
but the color is irrelevant. A model of a wall color must have the same color,
but can be any material or size.
i. Model Other Specific Situations
If you want to know if some specific drug will cure a disease in humans, you could test it in an animal as long as the animal reacted like a human to the drug.
ii. Model as an Instantiation of a Fundamental Process
Alternatively, you may not be so much interested in modeling a specific situation but rather all situations. For example, rather than seeing simply pecks to the blue light at 90% body weight, you may see the behavior as an instantiation of the effect of altered motivation on behavior as envisioned within some paradigmatic structure. Because science is the quest for general knowledge, virtually all research is seen as a model of a broader process. In sum, we are not so much interested in discovering some unique finding, but rather we are interested in discovering the common laws that explain the unique finding. We want to know why things are the same.
When studying behavior, two distinctly different kinds of questions emerge. One type asks things such as, “How fast can a pigeon peck,” or “How many colors can pigeons discriminate?” A second type, one that's vastly more important to psychology, asks things such as, "Why does this type of experience produce that type of rate change,” or “What type of experience produces that type of control by the stimuli?” Note that neither of these questions necessarily requires a reductionistic explanation. Whether the explanation appeals to higher or lower levels of molarity or shorter or longer time scales is a different issue.
This is the specification of how fast a pigeon can peck or the sensory capacity of a cat or the memory span of a person. This type of research is focused on determining the pattern in the behavior or the capacity of the organism. This would be the determination of how much convergence was necessary before a person detected depth in the image or the pattern of walking used by various insects.
This is the specification of the functional relationships causing behavior to be what it is. The difference between the following categories is basically a matter of the realization of what's going on. “Static” is only to those who do not see change, “dynamic” and no more is only to those who do not understand the underlying relationships. “Relationship” only, is to those who cannot specify a thing’s controlling factors. “Specify” only, is to those who do not know the paradigm within which the specified relationship fits. Each of these descriptions can be applied to the same research task. It’s just what the researcher understands. Obviously, to do any of them requires some rudimentary understanding of what factors are relevant to describe.
a. Describe a Behavior’s Static Properties
This activity is simply describing an observed behavior without describing its changing aspects. Clearly behavior changes as a function of many variables, especially time, and the meaning of the behavior is with respect to the absence of that behavior. However in this case, the dimension across which the behavior changes is not systematically noted, e.g., the kangaroo jumped, or John slept.
b. Describe a Behavior’s Dynamic Properties
This activity describes the changes in an observed behavior without describing in detail how that change is related to other nontemporal factors. Very often this type of observation is overlooked. For example, as John sleeps many systematic changes take place over time, the most well known being entry and exit from REM sleep.
c. Describe a Behavior's Relationship with Covariates
This activity systematically describes the changes in both an observed behavior over time and changes in other events which vary with the behavior without extensive examination of the necessity or sufficiency of that covariance (e.g., the distance of a kangaroo jump is a direct function of the number of dingoes present and an inverse function of their distance).
d. Specify a Behavior’s Controlling Factors
This activity describes changes in behavior and how that behavior is a function of some controlling factor without precisely integrating those controlling factors into some relatively complete, paradigmatic context. This type of research strives to determine the environmental determinants of the behavior, especially the consequences maintaining the behavior. The isolation of track convergence in the railroad track/depth perception example.
An essential aspect of this activity is the separation of various potential causal factors and their independent manipulation. For example, if it is believed that the apparent convergence of railroad tracks makes them appear three dimensional., then looking out a window at railroad tracks can be compared to looking at two-dimensional pictures with tracks converging and not converging for their appearance of depth. The separate manipulation of “real” depth (the window) and the depth “cues” (pictures with and without convergence) is an essential aspect of this research. Without the separation of the cues and the reality, you would be left with nothing to say except that apparently the person knew depth. A “researcher” with no aversion to tautology could then claim that it must be that the person has a depth understanding center in their brain and that’s how people react to depth correctly. Another “researcher” equally unintimidated by tautology could claim that it must be that the brain evolved with the ability to know depth because it was reproductively successful.
In this sense, research must show how the animal is “stupid,” rather than show how the animal is “smart.” Showing a person real depth and measuring the fact that they know real depth gives us no knowledge. The only explanation available is the postulation of an inner process. If you show that the subject is stupid (i.e., sees depth in a two-dimensional picture), then you have discovered the cause. This is the meaning of the quip “stupid animals prove researchers smart, smart animals prove researchers stupid.”
A modulating influence is a factor which can alter the occurrence of a behavior but which does not necessarily "cause" that behavior to occur. While hitting oneself on the thumb with a hammer may be seen as the cause for screaming out; the words selected to articulate the exact nature of your feelings are modulated by the people near you. A previously well-established effect is modulated as a result of the new procedure. Obviously this type of manipulation provides information on both the variable causing the behavioral modulation and on the behavioral process being modulated.
The only way to find out whether or not one event truly affects or "causes" another is to systematically change or manipulate the occurrence of the supposed cause and see what happens to the effect. Unless this is done, a causal relationship between the two events has not been demonstrated. Just because some child started to walk after using your snake oil therapy for the first two years of its life does not give you the right to claim that the therapy should get the credit. You must actively manipulate the independent variable while holding all other variables the same if you wish to infer causation. You must give the therapy to some children and not to others. If only therapy children walk then the therapy can claim credit for the ability to walk.
The following terms relate to the notion of causation. Each captures some element of the concept, but as you can see, their specific usage is not without some problems. The inference of necessary and or sufficient causation implies some frame of reference. If the frame of reference is allowed to change without limit then establishing either is meaningless. In establishing light as sufficient to elicit pupillary contraction in an animal it is presumed that the range of conditions does not include overriding drug conditions or living and dead animals. A necessary cause can similarly be so depreciated by variations of its interpretation that it becomes virtually useless as a description. Aristotle's enumeration of causes, raise other issues. While we tend to think of a single event like the woodsman's chop as the cause of a tree falling, we could alternatively point to the role of the tree being there and gravity as causes.
A sufficient cause is one which is capable of causing an event to occur but which is not essential to its occurrence. A match, candle, and spark are each sufficient to ignite gasoline but none are necessary. Light is sufficient to elicit pupillary contraction, but it is not necessary; drugs and emotional state can also contract the pupil.
A necessary cause is one which is essential for an event to occur. Oxygen is essential for gasoline to ignite and burn, however it is not sufficient. Life is necessary for pupillary contraction, but it is not sufficient.
The component parts from which something comes, such as the raw materials of a tree. earth, air, sunlight and water.
In psychology the material cause of a behavior is its prior history (including evolution) or its underlying physiology from which the functional relationship is built.
The propelling factor which sets it into motion such as the woodsman's chopping causing the tree to fall.
Effective stimuli - the simplest example would be an eliciting stimulus.
e. Integrate the Functional Relationship within a Larger Framework
This activity provides a general well integrated framework or paradigm within which an event can be understood. The controlling factors for a behavior may be correlatively specified in terms of its evolutionary, developmental, learning, and perceptual context. In this sense, description sufficient to provide prediction detailed enough to be labeled causation within a broader context is an explanation. This aspect of science is crucial because it provides the broadest base of support for the phenomena. It is the antithesis of eclecticism. Eclecticism is, by definition, without any integrating systematic framework. The therapist or researcher picks and chooses therapies or behavioral processes willy-nilly for no other reason than it appears to serve their purposes in that situation.
Simple eclecticism is the professional sounding word for ignorance. It is ignorance because there is no specifiable way to reliably choose a course of action. See the beginning of Chapter 2.
This aspect of research proceeds by breaking a phenomenon down into simpler elements. Analysis is based on the assumption that the action of a whole is the result of the action of its parts and their interaction. By isolating the parts and coming to understand their simple processes, then complex wholes can come to be understood. The belief is that the complexity and unpredictability of wholes is due to the action of the many small difficult to control processes making up the whole. Analysis is specifically designed to obtain information concerning the nature of the underlying behavioral process by breaking the phenomenon into its parts. This is the process of isolating active variables or ingredients, or the removal of irrelevant or confounding variables. Example: If given boxes and a hanging banana, a chimpanzee will move the boxes to form a ladder and will get the banana. We can easily show that that activity is not some mystical or transcendental insight by using analysis. By providing or withholding various component experiences, we get predictable variations in the final behavior. Experience with each precursor is necessary for the complete behavior to emerge.
An extremely important realization for a researcher to make is that the task is to show why the behavior occurred as the result of simple environmental experiences by proper analysis. To show that the behavior had to be the result of a “smart” animal -because you were unable to isolate the cause- is to have failed as a researcher: The question “why” has not been answered. Thus, “... and then a miracle happens,” or “... and then the animal realized the right solution,” or “... and then the animal used its cognitive map”-type of research is pointless. Rather than uncovering a cause for the behavior, the researcher needlessly demonstrated once again that sometimes animals do things that appear very “intelligent.” We already know that. The point of research is to discover why. In perception research, we need not demonstrate that people know that some objects are far away. We already know that. We need to determine what aspect of the stimulus makes the person react as if the object were far away.
A related error is to assume that a plausible excuse is an actual explanation. Often the discussion section of a paper offers a plausible excuse for the obtained effect without the empirical analysis required to convert the “likely story” into a “proven fact.” These plausible excuses often seep into the literature as “proven facts.” Much time has been wasted by researchers assuming that reasonable sounding explanations in a discussion section were factual explanations proven in the results section.
This class of analytical experiments analyzes a relatively complex behavioral phenomenon by breaking it down into each of its proposed component processes for an eventual correlative description and subsequent integration into a coherent set of laws.
This class of analytical experiments is to see if a theory is true. The theory is “analyzed” by formally or informally deriving predictions from the formal or informal theory and testing them. The research is not at all interested in what a particular subject will do in a particular situation. Rather the research is to examine the theory's ability to predict correctly at the outer edge of its applicability. If theory A is correct then the behavior will occur one way whereas if theory B is correct then the obtained behavior will be different in some obvious way (using the steam shovel under the lake metaphor, this would be testing whether a series of high spots (boom) came out of one end as the theory would predict. Note that you don't really care about the high spots other than as support for the theory which suggests that it is a steam shovel down there.).
It is occasionally asked "why study pigeon pecks?" or "who cares why a pigeon pecks?" Both of these questions indicate a fundamental ignorance of what is being done. The problem is easy to illustrate by changing the research area to personality and asking "why study people putting pencil marks on an answer sheet?" Another example would be “why look at squiggly lines” when the actual discussion is about the image in a random-dot stereogram. In all three cases, the error comes from a failure to understand the abstract nature of the research question.
Synthesis is the putting together or creation of something. It is an important stage in the empirical collection of knowledge because it provides feedback with respect to the validity of the presumed process. If you are collecting information correctly and in such a way that you understand it, then you can generate correct theoretical models of the presumed underlying process and you can create or synthesize new forms of the behavior at will.
The analysis phase is the first stage in the construction of an integrated framework of explanation. Synthesis is the next stage. The synthesized results demonstrate the validity of their presumed causal mechanism. The purpose of synthesis is to assemble known parts into a whole. The result is the production of a complex behavior or an integrated theory.
This type of synthesis is the production of a specific behavioral phenomenon, specifically as a test of your understanding, or for some practical use (which also validates your understanding).
This approach proceeds by reconstituting the phenomenon from component parts in order to identify the underlying cause or to assure that the analysis which specified its component parts was correct.
Example: The perception of orange is thought to be mediated by yellow and blue receptors. By illuminating a panel with monochromatic blue and monochromatic yellow you determine if orange is a simple combination of blue and yellow.
ii. Modulation via Necessary Element
This procedure demonstrates the causal factors underlying a phenomenon by causing it to occur and to cease by introducing and removing a necessary element.
Example: You come to believe that the amount of your roommate's studying is under the control of the social attention you provide. You test this by providing and not providing social attention on random days while you monitor the amount of studying.
iii. Demonstration with Other Subjects or Species
This procedure demonstrates the causal factors underlying a phenomenon by demonstrating the phenomenon in animal models.
Example: Animal models could be used to demonstrate that language evolves and works the way we think it does.
b. Theoretical Synthesis or Paradigmatic Integration
This is the creation or synthesis of an explanation or theory. It is the integration of an event into a large coherent body of knowledge. A detailed treatment of what it means to explain a phenomenon is covered later in this chapter. Theoretical synthesis could be seen as essential for the establishment of a phenomenon’s complete cause in the Aristotelian meanings of the term "cause."
i. General Theoretical Model
This type of synthesis is the specification of how the phenomenon is thought to be caused and how is it thought to work. It is the model. This was what Aristotle referred to as a formal cause. It is a thing's implicit form or meaning through precise metaphors and models. The laws underlying or describing the action. The model can be reductionistic or correlative.
This type of synthesis demonstrates the correctness of a model or theory for something by making correct predictions based on those factors. It is understanding the cause because you make correct predictions. Predicting that the marble will fall through the hole in the pan. This was what Aristotle referred to as final cause. The specification of the end or terminal state of something. An example would be the hydrodynamic efficiency in a shark. Evolutionary adaptiveness, functional goals, and equilibrium solutions are various forms of "final causation." Potential problems with these kinds of conceptualization are discussed under the section on teleological explanations.
iii. Integration of Divergent Phenomena
This type of synthesis demonstrates the correctness of a model or theory for something by showing how that theory can make sense out of numerous phenomena which are otherwise intractable to explanation.
Example: It was shown that the explanation for credit card debt, flunking out of school, unwanted pregnancy, and substance abuse; were all understood with the discounting function.
Send comments/criticisms/speculations to
Date Last Reviewed: November 17, 2002