When one receives recognitions such as the Society of General Psychology award, one is obviously pleased and proud. I thank the society for bestowing this honor; the society has been a home to me for decades. But the honor never could have happened without many people playing important developmental roles over many years. In my case, these teachers who groomed me and packed me off to college, professors who provided stimulation and opportunities and important words of calm and wisdom, my own students who creatively and vigorously carried out our shared research ideas, and generations of psychologists before us that established the educational, scientific, and formal organizational contexts within which it was all possible. Each of these in different ways are critically important to one's success, and I thank them all. Without any one of these, including an ever-patient wife, there would have been no career.
The first lesson here—the first of several I mention herein—is that people are essential to your successes. Ideas may be important and constructive, but people are the key to unlocking the future.
So, what does one talk about given this honor. It really is a dilemma. I chose to talk about some lessons I have learned in hopes that they will be informative to others. The talk was filled with personal anecdotes provided in historical context. This will remind you of the “old days”, yet hopefully each has a kernel useful even in modern contexts.
I want to comment on how it was that I got into psychology at all. I was at Kenyon College, a small liberal arts college in Ohio. My major was chemistry, and I have a BA in chemistry. In Spring of my junior year, I was dating a girl at nearby Denison College who was a Psychology major. She dragged me to a couple of Psychology Department colloquia at her school. One lecture was on the then famous “Executive Monkey” experiments with paired monkeys that received matched electric shocks but differed in the psychological context of the shocks. The experiments revealed that psychological factors surrounding stressor could result in fatal gastric ulcers. The second lecture was on Korean War POWs. The Korean War POWs were of special interest to the US military for the so-called brainwashing they received, their failures to escape, and for the many instances of young men simply lying down, pulling a blanket over their heads, and being found dead the next morning. This latter was attributed to the unpredictable treatments of abuse and reward and the social isolation created. They just “gave up and died”.
Together these lectures changed my life and guided some of my research foci for the future. The message I took away from both of these lectures: “The mind can make you ill and even can kill you.” Now, that is worth studying. I was sold. I applied to graduate school in psychology. These two studies also planted the seeds of interest in consequences of unpredictable and uncontrolled aversive events…seeds that sprouted later in my career as learned helplessness and psychological factors in stress and gastric ulcers. Why do I discuss this here? Not for my career reasons, but as guidance to you students, faculty, and practitioners. This anecdote illustrates the importance of a single lecture or two.
The second lesson here is addressed to faculty, students, and leaders: When you are invited to give a lecture or colloquium someplace, say yes! You could change somebody's life. When as a young scholar, you have a chance to go to a lecture or colloquium—even if the topic seems oblique to your current interests—go. It just might change your life. Chairpersons, please include adequate funding for external colloquia in your budgets. Colloquia are important educational fare.
Now let me talk about a surprising lesson from graduate school. Graduate school was a challenge. I well recall my first week at University of Pennsylvania and my introduction by Richard Solomon to his animal laboratory. I had never worked with animals before. After a two-hour tour of the laboratory, I was handed a key and told, “Your publishable manuscript is due in May” OMG. I was terrified. I knew nothing.
Well, the project I developed and carried out tested OH Mowrer’s implicit hypotheses on the nature of anxiety and the development and persistence of phobic behaviors. The key feature of this “two process” analysis was that “phobias” were essentially avoidance responses that developed to solve the problem of anxieties. Avoidance responses were thought to be the interaction of two separable learning processes: Pavlovian conditioning of fear/anxiety and acquiring an instrumental response to cope with the anxiety, that is to “escape” the anxiety.
The project set out to explore the separability and sequential independence of the instrumental and the Pavlovian components of the phobic-avoidance behavior. Specifically: were effects of discriminative Pavlovian fear conditioning upon avoidance responding the same independent of which process has taken place first. And, was there an order effect? The project did get done, although it did not get published by June, although it eventually did get published. Mowre’s hypothesis was largely confirmed. In either order, the Pavlovian fear signal strongly and about equally evoked the instrumental response. But there was an order effect that I wanted to follow up, more about this later. The reason for bringing up my first year project was because—like all graduate students—I saw the challenges before me of that first year at Penn as an impossible one.
The third lesson here for young scholars is that the faculty are your benefactors. They want you to succeed. The “impossible” challenges they give to you are not impassable obstacles, but stepping-stones to your success.
Moving on, I came to experimental psychology of learning in 1960 just when Throndikian S-R Behaviorism—wherein Stimuli (S) get associatively bonded directly to Responses (R)—was transitioning to so-called Neo-Behaviorism. Basic Behaviorism actually was much more powerful than it is credited in modern texts and discussions. And, Neo-Behaviorism even more so. And both provided formalistic anchors for cognitive-type constructs that otherwise are left free floating even today.
Behavioristic theories challenged its proponents to test the S-R theory and its versions; Hull's theory was wonderful for this because it made very specific data predictions. And, it was constructive even as it proved itself wrong. Behaviorists found that they needed a way to incorporate the concept of goals into its structure. Neo-Behaviorism discovered new phenomena that required invoking “mediating states.” They did this by invoking “conditioned fractional anticipatory states” to explain some motivation-like and frustration-like effects. These mediating states were assumed to arise out of Pavlovian S -SR relations embedded in the S-R-SR events and allowed them to retain the general associative analytic structure.
The embedded S -SR relation as “CS-US” comes to evoke a “fractional anticipatory goal response” of (rg- sg) as mediating state such that R evocation is two steps or links:
S → (rg- sg) → R → SR Goal → (R to Goal rg and its Stimulus feedback sg),
LINK 1 LINK 2 joined by the conditioned (rg- sg) as a mediator.
So, in this theoretical approach there were two associative links based on S→rg and sg→R that could be analyzed and independently manipulated and the mediator quality was influenced by choice of the reinforcer outcome events. As this sketch implies, over the years, I think we did nearly every imaginable variation possible in what were called Pavlovian-to-Instrumental Transfer experiments.
Inserting a mediator raised questions as to, what were the behavior-modulating properties of these mediators. As an example, one could contrast different theories within Neo-Behavioristic approach about the response controlling properties of the mediator. Was the mediating property as a motivation or a cue? Most following Mowrer argued that the mediator was general motivational and non-specifically energized action. Trapold and I argued the mediator was very specific to the specific reinforcer, thus incorporating goals into the behavior chain, and these anticipations of goals had cue properties that guided responses. This very question led to a long series of published experiments in our lab and with collaborators on discriminated conditional choice behaviors. These yielded phenomena called the Differential Outcomes Effects. These differential outcomes effects revealed new phenomena that garnered substantial attention including faster learning of discriminations, persistence in the face of otherwise disrupting factors like drugs, improved short-term working memory, and enabled learning by children and college students of relations that were otherwise unlearnable.
The fourth lesson here is that nothing is so useful as a well-articulated theory. Even when wrong, a good theory can stimulate new approaches, new research and new findings. And, therefore, do try to work within a theoretical context.
My PhD mentor, Richard Solomon, was interested in adaptation to hedonically important events such as pain and even rewards. NIMH supported his major research efforts on this. Students working in Solomon’s Laboratory Group were focused on adaptation to pain, learning of fear, as well as testing aspects of Two-Process Theory. In this project, animals were exposed to long sequences of uncontrollable aversive electric shocks. Of interest was the change in heart rate prior to, during, and after each shock and how these changed across the series. These changes later showed up as foundational for Solomon & Corbit's 1974 Opponent Process Theory of Motivation. This theory derived its empirical basis in part from these experiments on sequential shocks.
This Opponent Process theory of affective dynamics was actually patterned after the Hurvich-Jameson opponent process theory of color vision. In any case, the theory accounted for the observed dynamic changes in affect across serial exposures by hypothesizing an unchanging primary response (a) and an accumulating experience-based opponent response (b) that grew with each exposure. The Opponent Process Theory of Affective Dynamics is another excellent example of how a carefully articulated theory can stimulate research and guide practice.
But back to the basic experiments on exposures to a series of painful events. I mention these ongoing experiments because they afforded me a serendipitous opportunity to re-explore my first-year project effects of prior uncontrollable shocks had upon animals’ ability to later learn avoidance responses. As animals completed the experiments in which they were exposed to the series of uncontrollable shocks, I asked permission to test their ability to then learn avoidance responses in a shuttlebox. The results were dramatic. Animals tested the next day simply did not learn to avoid. Not only did they not learn to avoid, mostly they did not even try to escape. In contrast, previously unshocked control animals immediately responded to the shocks by running and escaping. I thought back to the executive monkeys and the Korean POWs.
This “after-thought” supplemental testing for escape and avoidance learning was the first of the “Learned Helplessness” series of experiments that we graduate students—Seligman, Maier, and I— explored further together. The primary features of Learned Helplessness were affective unreactivity and inability to learn from those rare occasions on which they made an escape response. After Seligman and associates later argued that Learned Helplessness was a model for depression, many researchers became very interested. Learned Helplessness became a research and theoretical hot-topic for many years (Google Scholar says 17,000 papers have LH in the title). But it was almost serendipitous, and it was done by students in graduate school.
The fifth lesson is that young scholars and even graduate students can and do make important discoveries. Sometimes such discoveries are not preplanned.
Had we been required to “pre-register” our experiment and design—a common theme in contemporary psychology—I would likely never have made the discovery. That is, while “pre-registration” may be a positive thing for some classes of replications and advanced tests of theory, it may be a barrier to serendipitous findings. Wondering “What if?” is sometimes a good motive in science. By the way, we students thought the phenomenon was cognitively based while Solomon thought it motivationally based. But Solomon let us pursue our ideas anyhow. Now, 50 years later, work by Maier suggests Solomon was right.
Does one always know what is important? A major mid-century question was “How do animals learn avoidance responses?” The traditional training procedure, harking back to Neal Miller and Hobart Mowrer in the 1940s was to present a warning signal of the impending aversive event and if the animal did not make the required response during the signal, the aversive event was turned on and continued until the animals made the response to escape the aversive event.
These opportunistic escape responses were thought to teach the animal what was required. Avoidance responses were commonly characterized either as an extra-fast escape response (Hull) or as an “escape” from an acquired fear of the aversive event (Mowrer). There was even a1948 experiment by Mark May that purported to demonstrate that avoidance responses were basically “escape” responses.
It seemed to me that resolution of this debate about avoidance as a class of response acquired through escape was important. After all there were dozens of treatises on the topic well into the 1970s.
In our lab in Minnesota, we had already shown that presentation of a separately learned CS for aversive events could when presented to an animal evoke a previously learned traditional avoidance response. But of course, the avoidance response had been trained in the traditional way which meant the animal had first learned to escape the aversive event and then learned to avoid it. We sought to unconfound escape and avoidance in the basic instrumental task.
First, each of two groups of animals received Pavlovian conditioning in which a CS was paired with aversive shocks. We then trained two groups to make the critical response, but we trained them in different ways. One learned the required response in a pure escape task. The other learned the required response as a pure avoidance response that did not allow for escape. Finally, we presented the Pavlovian CS+ cue while the instrumental response was possible. The Pavlovian CS+ only elicited the instrumental response in the avoidance-trained group. It did not do so in the Escape-trained group. This meant that escape responses and avoidance responses were fundamentally different.
We published, of course. In those pre-internet days, one ordered hard-copy reprints to send out to people who wanted to have copies. I thought this was a wonderful result; many people would find this important as resolving the question of the nature of the avoidance response (and its dependence on escapes). I ordered reprints. One usually ordered 50. I ordered 300. Recently, as I closed out my office and files, I found those old reprints. I had 295 left. And, one of the five sent out went to my mother.
The sixth lesson here is that what you think is clever and important may not be viewed as such by others. Nonetheless, such research can be informative to you and how you do future experiments. Not every experiment you do will be a break-through. You must be resilient and persist.
Is Psychology in Crisis? This topic is quite multifaceted and well beyond our dealing with it in a deep fashion. But I do have some opinions that I would like to share.
One goal of science is the discovery of new reliable principles. Discovery sometimes arises from simply “asking what if” and then exploring the data. Alternatively, it may involve making a bold hypothesis and designing a novel experiment to test it. In our lab, we used both approaches—sometimes successfully, sometimes not. One rarely knows how an experiment will come out, that is why we call it experimentation.
A current trend is to promote “pre-registered” experiments. Part of the reason for this is to address some of the "supposed crimes" in data analysis. These crimes go by a variety of names, such a “p-hacking” and HARKing (hypothesizing after results are known), among others. Pre-registration constrains one to propose exactly what one is going to do operationally and analytically.
As I have already illustrated, with the example of Learned Helplessness, pre-registration can limit the likelihood of novel discovery. Not a good thing in our business, in my opinion.
We must distinguish between two types of research: Exploratory and Confirmatory. Pre-registration is, I agree, a useful tool for experiments attempting replication of a work—especially if the original investigator gets to comment on the proposed pre-registered experiment to ensure faithful capture of the original work. Failure of the pre-registered design and methods to faithfully capture some part of the original results in the new work being a “systematic replication” rather than an exact replication. Now, systematic replications are themselves useful because, if successful, they extend the generality of the original experiment. And that is useful. But if unsuccessful, they do not refute the original work and cannot be called a “failure of replication”. Results from our laboratory typically have reliably withstood both replications and systematic replications.
Based on our different interpretation of the nature of mediators in response selection, we trained subjects in a discriminated conditional delayed choice task. For one group, the different correct choices all resulted in the same reinforcer. For the other, the different correct choice response each resulted in a unique reinforcer. The result was discovery of a new phenomenon, called Differential Outcomes Effect, as noted earlier. This phenomenon turns out to be a broad and very powerful effect influencing many aspects of learning and memory, and it does so across tasks and species. There are data that suggest that this change in how to reward choice responses actually changes how the response-choice relation is stored neurologically. I know of no exact replications, but I think they are unnecessary because of the many systematic replications.
This seventh lesson is as pre-registration goes, we need to distinguish between Exploratory Research and Confirmatory Research. And within Confirmatory Research, distinguish between systematic replication and exact replication. Each has its place
Not everybody agrees that there is a crisis of replication or p-hacking in psychology (or in science more broadly). They see successful replications and non-replications as the normal process of science. Because of our level of ignorance about the brain and behavior, we can expect most of our hypotheses to be wrong. The use of standard statistical procedures means that some portion of these wrong hypotheses will get statistical support…and perhaps get published. Some refer to this as the base-rate problem.
What I see as the “true crisis” is that “so-called scholars” go right on citing the claims that have failed in multiple replication tests. Sadly, some authors do continue to cite papers that have failed replication. When scholars do this, they promulgate error. So, in at least one sense, science is not self-correcting or at least not efficiently so.
I am not a statistician. And I am not especially fond of statistics although, of course I do use them. I would rather not. It cannot be said too often: “Statistics are not the results.” Rather, statistics are a heuristic for thinking about the results of experiments. Tukey’s wonderful book, Exploratory Data Analysis makes this clear.
If you can avoid using statistics, do it. I sound like a Skinnerian, and there is much to laud about that approach to data. My favorite statistic is what I call the “inter-ocular trauma” test: If the data smack you between your eyes, they are probably meaningful. Shortly after David Grant became the new editor of the premier Journal of Experimental Psychology (JEP), he published an editorial saying that from then on. all papers published in JEP had to include their ANOVA tables. I could not resist. I submitted a paper that had no statistics and hence no ANOVA. I did show the data for every single trial for every individual's response to each of three test stimuli in a graph—some 360 data points. It was obvious that the animals responded promptly to the CS+ and responded only the CS+. They did not respond to the CS- nor to the SD for the trained response which had been extinguished before the Pavlovian conditioning. My results section began: “These differences are so large and so consistent as to render statistical analyses superfluous.” To my surprise—and his credit—Editor Grant agreed.
Nonetheless, most exploratory research requires exploratory statistical examination of the data for various features. Not just a single pre-determined pre-registered statistical test.
So, the eighth lesson is that performance data are the results of your experiments. Not the statistics. Replication efforts are critically important in science. However, failures to get the same statistical outcome can arise for many reasons. Statistics are merely a heuristic for thinking about behavioral data. Graph your data and use the intra-ocular trauma test. Think twice about citing a study that has failed a good faith replication.
Although I was as a student unaware of the hierarchy of how psychology is organized in the world, I quickly came to see that there was a set of structures that made the discipline of psychology work. Yes, you have an experimental result, so what next? Present at a formal departmental meeting, national or international meeting, or publish in a journal? Those meetings, journals, and grants only exist as elements in truly huge organizational structures. And you and your work are dependent upon them. Many people make up the world of psychology: researchers, academic teachers, clinical practitioners, and those applying psychology in industrial/organizational/military contexts. Those sustaining the system as reviewers, editors, organizational officers maintain the systems that sustain it all. They have been doing so for generations and are needed to do so into the future. Yet, we all describe ourselves as “psychologists". We use the work of our predecessors to make it possible. These much-needed structures are in place to make our futures as psychologists possible.
The world of psychology is bigger than you might guess. A million of us spread across the globe. We say we are psychologists, yet we may be very different. In the USA, you generally need an advanced degree. In Vietnam, you need two college level courses. So, we may ask, “Is psychology a unitary discipline?” Physicists or chemists have unitary disciplines. They can step into any lab or facility in the world because they share common knowledge. Currently, this is not so worldwide for psychologists. This challenge is the focal issue of organized psychology.
As members of our state, regional, or national association or our state association, we send representatives to the U.S. National Committee, which is part of the U.S. National Academy of Sciences, which in turn support U.S. representation at the International Union of Psychological Science, which in turn has representatives at the International Council of Science (ICSU), headquartered in Paris. The International Science Council organizes international research projects in countries around the world and also sponsors "International Years", such as the International Geophysical Year, which involve research from all of the 30 international science unions. The ICSU projects provide reports to guide UNESCO. It is a long chain that functions on your behalf. However, to function well these organizations need you to volunteer your efforts to manage the organizations and make sure they know what you and I—“WE”— want and need.
The ninth lesson is that for psychology to function and function well at the local, national, and international level; we—you and I—have to give our support and efforts to organized psychology. We have a moral responsibility to do so—to give back to the discipline that has been good to us. Sadly, neither the nature of our organizational structures nor the moral responsibility to maintain them is taught in college.
Unlike many who focus narrowly and make major contributions, I have been an intellectual dilatant from the classroom, to laboratories across the globe, to participating in all the organizational structures of our discipline. But I tell you, it has been a world of fun. I thank all my past students and colleagues for that. Herein, I have looked back over my career and noted some lessons I have learned. I have put these into context and shared them with you. I hope that you, professionals and students, found at least some worthwhile nuggets here.