I was walking across campus one day when I saw Ingrid crossing the sidewalk a hundred feet or more ahead of me. She had been a prize student in two classes, so I gave her a hearty wave. She smiled and waved back, but some infinitesimal discrepancy in her behavior suggested that something was amiss. Maybe she was harboring smoldering resentment from the semicolon incident. I couldn’t say what, or why I thought so, and I forgot about the incident until months later, at graduation, when I discovered that she had an identical twin. I had obviously hailed the twin who, with the ennui of twins everywhere—”Yes, I know, I know. I look just like her.”—merely smiled and waved and carried on with her day. Seeing them, side by side, in identical caps and gowns was a delightful surprise, a thrill.
As an aside, this anecdote illustrates how exquisitely subtle our discriminations are of the behavior of others. How can we quantify the difference between a polite smile and a smile of recognition? How short must the latency of a wave be to forestall suspicion that something is wrong? How difficult, then, is the task of the behavior analyst who wishes to manipulate social reinforcers in applied settings! Does “Good job!” come across as approval or as the well-practiced routine of a bored behavior specialist? What are the stimulus properties of the difference?
But to return to my main point, when we look from one person to another, our nearly invariable experience is one of discrepancy, so seeing identical twins is surprising, arresting, and we respond with special force, a “force” that has stimulus properties. So it is with all coincidences. I am speaking here, not of correlations, like the size and weight of a can of beans, but of unpredictable events that share salient stimulus properties. Coincidences jump out as us, as in the following anecdote about schoolboy Skinner:
In the grade school that I attended as a child, a single teacher taught two grades in the same room. While one class recited, the other worked on its assignments. One day in third grade, when my teacher was talking with the other class, I raised my hand, waved it wildly to attract her attention, and said “I was reading the word ‘middle’ just when you said it.” Both classes laughed. I had been impressed by the coincidence, but I should have been impressed by the fact that I was impressed.1
The title of Skinner’s paper was The Force of Coincidence, and it aptly provides the text for my point: Coincidences have “force.” That is, they have relatively strong effects on a particular topography of behavior, which we detect as a discriminable jump in response strength (e.g., latency, magnitude, or those incipient response tendencies we tact as “on the tip of our tongue.”). I am proposing that the discriminable jump in response strength was itself surprising, surprising enough that 8-year-old Skinner waved his hand wildly to announce the coincidence to the class. The controlling variable for the announcement was not the textual stimulus “middle,” for it could plausibly have been many other words; nor was it that particular compound stimulus, for we can’t suppose that it had been encountered on a previous occasion. Rather, it was the force of coincidence on his repertoire that evoked his response. He was tacting a private event, the surprising effect of a coincidence on his own behavior.
The force of coincidence is plausibly the controlling variable in match-to-sample procedures: A sample stimulus is presented to which the subject is assumed to respond, or is required to respond. (And the subject who does not respond will be unable to match when the comparisons are presented.) Then an array of stimuli is presented, only one of which evokes a response like that to the sample. Pecking it, pressing the lever under it, pointing to it, tacting it, or clicking the mouse on it will lead to the delivery of a reinforcer. A child who has been praised for pointing to a triangle in the presence of a sample triangle, might have learned one of three things: Always point to triangles, always point to triangles in the presence of a triangle, or always point to whatever pattern corresponds to that of the sample. Only the third condition will satisfy the experimenter, and training is continued until such performance appears. When experimental subjects can reliably match a novel stimulus, we can assume that they are responding, not to the sample and comparison stimuli themselves, for those stimuli are putatively novel from one trial to the next, but to the force of coincidence. There is no history of reinforcement for responding to the raw stimuli in such a context, but every reinforcer is delivered in the presence of such a stimulus event.
Barry Lowenkron coined the term “joint control” for behavior under control of the force of coincidence, and it is a powerful and sufficient explanation for much complex human behavior. In a match-to-sample task, glancing from the sample to a matching comparison stimulus is like glancing from one twin to another and requires no special explanation. When a substantial delay is imposed between the offset of the sample and the onset of the comparison stimuli, there are no coincidental stimuli, and correct responding must be explained in some other way, setting aside magic. An obvious possibility, confirmed by Lowenkron’s research, is that verbal humans in such tasks mediate delays through verbal rehearsal, or through some other mediating response. Doing so would transform the coincidence from a visual one to, say, a verbal one.
Being appropriately cautious, Lowenkron did not define joint control in so speculative a way. Rather, guided by his experimental data, he defined it as responses under the combined (“joint”) control of a tact and an echoic response, both of which might be covert. But I see no reason to avoid the generalization that correct delayed matching to sample, or symbolic matching to sample, is mediated by the force of coincidence in any form of response.
But stimuli need not literally “coincide” to have force. We need not rehearse every instant during a delay to choose a matching stimulus. To the contrary, if a stimulus evokes a highly unusual response, we can detect a match after quite long delays in the apparent absence of relevant mediating behavior. My students and I have come to speak of this as the “Disraeli effect,” after an offhand comment I made in a paper on joint control. Disraeli was the Prime Minister of the United Kingdom in the 1870s, approximately at the peak of its power. That’s news to most Americans, who seldom follow the succession of Victorian prime ministers in any detail, but the name is distinctive and rolls off the tongue in an engaging way. Now the point of interest is that tomorrow morning when you are driving to work, suppose you pass a road sign that reads, “Disraeli Boulevard.” Will you not cry, “Hey! There’s that name again!”? But surely you would not notice signs for Power Road or Peak Place, even though “power” and “peak” were words in the sentence in which “Disraeli” appeared. Furthermore, if your next encounter with the name Disraeli occurs a year from now, you would be unlikely to respond with surprise or recognition. Some coincidences are more salient than others.
Coincidences have a variety of effects. The stimulus properties of coincidences are poised to facilitate responding in a variety of match-to-sample procedures because they are salient in the moment, but effects that endure beyond a minute or so presumably reflect conditioning. Conditioning is another function of “surprise” as research on blocking has shown for over a half-century, but that is another story.
1Skinner, B. F. (1925). The force of coincidence. In Reflections on Behaviorism and Society, Cambridge, MA: B. F. Skinner Foundation.