PEAK SHIFT

Introduction and Core Definitions of Peak Shift

The concept of Peak Shift represents a fundamental phenomenon within the study of learning and stimulus control, specifically observed following rigorous discrimination training. Primarily, it describes an alteration in the organism’s response pattern where the maximum frequency of response, or the peak of the generalization gradient, is displaced away from the stimulus that signaled non-reinforcement and towards a novel stimulus value that lies beyond the original reinforced stimulus. This unexpected displacement is crucial because it indicates that the organism is not simply learning to respond to the exact value of the reinforced stimulus, but rather is learning a relationship between stimuli. The definition of Peak Shift is traditionally bifurcated, encompassing its origin in classical conditioning experiments and its later application in theories concerning aesthetics and preference.

In its primary context concerning stimulus generalization, Peak Shift is defined as the movement of the response gradient’s peak in a direction that deviates from the less desired or non-reinforced stimulus, known as S-, to a point slightly past the value of the stimulus correlated with positive reinforcement, designated as S+. This displacement is counterintuitive because, if learning were purely absolute, the strongest response should occur precisely at S+. The observation that the maximum response occurs at a stimulus value that was never presented during training underscores the complexity of associative learning and highlights the influence of inhibitory processes generated during discrimination practice. Therefore, Peak Shift serves as compelling evidence supporting relational theories of learning over strictly absolute theories, demonstrating that organisms attend to the relative differences between stimuli rather than just their isolated properties.

The secondary, yet equally important, context in which Peak Shift is applied relates to aesthetics and the psychology of preference. In this domain, the phenomenon postulates that an exaggerated or extreme version of an already favored stimulus is often preferred over the normal or natural type of that same stimulus. This aesthetic application suggests that the underlying psychological mechanism that drives the displacement of the response peak in conditioning—the preference for an exaggerated feature—also dictates human and animal aesthetic choices, leading to a strong attraction to stimuli that surpass natural bounds in a dimension deemed desirable. For instance, if a specific coloration is deemed attractive, an exaggerated, hyper-saturated version of that coloration may elicit a stronger preference response than the natural coloration encountered in the real world, reflecting the core principle of displacement toward an extreme value.

Experimental Foundation: Discrimination Training and Stimulus Generalization

The empirical foundation for Peak Shift was established through rigorous experimental procedures, most notably in early studies of stimulus generalization conducted by researchers like Hanson in 1959, utilizing pigeons in operant chambers. The core procedure involves discrimination training along a measurable stimulus dimension, such as wavelength (color) or frequency (tone). Initially, subjects are trained to respond to a specific stimulus, the S+, which is consistently followed by reinforcement (e.g., food delivery). Following initial training, the critical phase involves introducing a second stimulus, the S-, which is presented intermittently but never followed by reinforcement. This S- stimulus is typically situated close to the S+ value on the stimulus dimension continuum.

The purpose of pairing S+ (reinforcement) with S- (non-reinforcement) is to teach the organism to finely differentiate between the two closely related stimuli, thus creating both an excitatory tendency around S+ and an inhibitory tendency around S-. Following this intensive discrimination phase, the subject is tested for stimulus generalization. During the generalization test, the subject is presented with a wide range of stimuli along the training dimension, including S+, S-, and numerous novel stimuli that fall outside the training range, all presented without reinforcement to prevent further learning. The frequency of the subject’s response to each stimulus is meticulously recorded to map out the generalization gradient, which graphically illustrates the organism’s tendency to respond to stimuli based on their similarity to S+.

The critical finding that defines Peak Shift emerges from this generalization gradient. Instead of the strongest response occurring precisely at the S+ value, the peak response is observed at a novel stimulus value that is slightly shifted away from S- and further along the continuum than S+. For example, if pigeons were trained to peck at a 550 nm (greenish-yellow, S+) light and not peck at a 560 nm (yellow, S-) light, the maximum pecking response would occur not at 550 nm, but perhaps at 540 nm (bluer-green), a color never previously associated with reinforcement. This consistent and robust finding challenged earlier, simpler models of learning that predicted the peak response would coincide exactly with the reinforced stimulus, compelling researchers to develop more sophisticated theoretical models to account for the observed displacement.

The Mechanism of the Shift: Absolute vs. Relational Learning

The discovery of Peak Shift played a pivotal role in the debate between absolute theories and relational theories of learning. Absolute theories posited that during conditioning, the organism learns the specific, isolated characteristics of the reinforced stimulus (S+) and the non-reinforced stimulus (S-). According to this view, the strongest response should naturally occur at S+ because that specific stimulus has the highest association strength with the reward. The observed shift, however, fundamentally contradicted this prediction, arguing strongly for a relational understanding where the organism learns the relationship between S+ and S-—specifically, that S+ is “greener” or “brighter” than S-.

Relational learning suggests that the animal learns to maximize the difference between the two stimuli. The displacement of the peak response towards a more extreme value of S+ relative to S- indicates that the animal is seeking a stimulus that is an even better example of the “correct” relationship established during training. Specifically, the animal learns not just “550 nm is good,” but “550 nm is good relative to 560 nm.” When presented with 540 nm during testing, which is even further removed from 560 nm than the training stimulus was, the subject responds more vigorously because 540 nm better exemplifies the learned relationship of being distinct from the inhibited stimulus. This mechanism highlights the active, comparative nature of associative learning, where inhibitory processes actively shape excitatory responses.

Furthermore, the degree of the shift is typically proportional to the proximity of the S- stimulus to the S+ stimulus. If S- is very close to S+, the need for fine discrimination is high, leading to intense inhibitory conditioning close to S+, and consequently, a larger displacement of the excitatory peak away from S-. If S- is far removed from S+, the discrimination is easy, the inhibitory gradient is weak, and the resulting Peak Shift is minimal or absent. This dependency on the training parameters solidifies the interpretation that the shift is a direct result of the interaction between the excitatory and inhibitory processes generated during the discrimination phase. The organism effectively optimizes its response profile to avoid the inhibited region while maximizing the distinguishing features of the reinforced region.

Characteristics of the Generalization Gradient

Understanding Peak Shift requires a detailed examination of the generalization gradient itself, which is a graphical representation depicting the strength of responding across a range of stimuli. In standard generalization experiments without discrimination training, the gradient is typically symmetrical and unimodal, peaking exactly at S+ and gradually decreasing in strength as the stimuli become increasingly dissimilar to S+. The introduction of discrimination training drastically alters the characteristics of this gradient, demonstrating the powerful influence of inhibition on learned responses.

The key transformation in the gradient post-discrimination training is the emergence of asymmetry and the subsequent displacement of the peak. On the side of the gradient closest to the S- stimulus, there is a pronounced steepening and depression of the response curve, reflecting the strong inhibitory control exerted by S-. Responses to stimuli highly similar to S- are suppressed due to conditioned inhibition. Conversely, on the side of the gradient moving away from S-, the response curve remains strong or even increases beyond the S+ value, culminating in the Peak Shift phenomenon. This asymmetrical shape is the visual manifestation of the conflict between the learned excitation centered at S+ and the learned inhibition centered at S-.

The width and height of the generalization gradient also provide insights into the underlying learning processes. A narrower gradient suggests highly specific learning, meaning the organism is only responding strongly to stimuli very similar to S+. However, the presence of Peak Shift indicates that the organism has generalized the response not based on absolute similarity to S+, but based on the relational distance from S-. The resulting peak, while strong, is situated at a value that was never directly reinforced, confirming that the inhibitory mechanism is not merely suppressing responses around S-, but is actively reshaping the excitatory tendency across the entire stimulus dimension continuum, yielding a response curve that maximizes the contrast between the reinforced and non-reinforced areas.

Theoretical Explanations: The Inhibitory Gradient

The most influential theoretical framework explaining Peak Shift is the Summation Theory of Generalization, proposed by Kenneth Spence in 1937. Spence hypothesized that discrimination training generates two distinct, overlapping gradients of responding: an excitatory gradient (E), centered around the reinforced stimulus (S+), and an inhibitory gradient (I), centered around the non-reinforced stimulus (S-). The observed response (R) to any given stimulus along the continuum is the algebraic summation of these two opposing gradients (R = E – I).

According to Spence’s model, the excitatory gradient peaks precisely at S+, reflecting the maximum association strength generated by reinforcement. Simultaneously, the presentation of S- generates a gradient of conditioned inhibition, peaking at S- and decreasing in strength as stimuli become less like S-. Since the S- stimulus is intentionally chosen to be close to S+, the inhibitory gradient overlaps significantly with the excitatory gradient. When these two gradients are algebraically summed, the inhibitory influence substantially reduces the response strength in the region between S+ and S-. Crucially, the inhibitory gradient also extends slightly past S+, causing a suppression of the response directly at S+.

This subtractive interaction causes the maximum response (the peak of the resultant R gradient) to be pushed away from the area of maximal inhibition (S-) toward the region where the excitatory tendency is still high but the inhibitory influence has sharply declined. The result is the displacement of the peak response to a novel stimulus value slightly beyond S+ in the direction opposite to S-. This elegant mathematical model successfully predicted the occurrence and magnitude of Peak Shift, providing strong support for the notion that associative learning involves the complex interplay of both excitation and inhibition, and that the final behavioral output is a net result of competing internal states. The inhibitory gradient, therefore, is not merely a mechanism for suppressing unwanted behavior, but a critical force in reshaping the entire landscape of stimulus generalization.

Peak Shift in Aesthetics and Preference

Beyond the controlled environment of the conditioning laboratory, the principles underlying Peak Shift have been extrapolated to explain phenomena in aesthetics, mate selection, and strong preferences, often linking the concept to the notion of the Supernormal Stimulus. When applied to preference, Peak Shift suggests that if an organism develops a preference for a stimulus feature (S+), an exaggerated version of that feature—a stimulus value beyond S+—will elicit a stronger affective or approach response than the original, naturally occurring stimulus. This application posits that the aesthetic preference system operates relationally, much like the conditioning system.

The classic example often cited involves sexual selection and human attraction. If a specific facial feature, such as large eyes, is associated with youth and attractiveness (S+), individuals may show a heightened preference for artificially exaggerated versions of that feature (e.g., in cartoons, makeup, or fashion models) that exceed the biological norm. The hyper-stimulus represents the Peak Shift, displacing the preference away from the biologically normal S+ toward an unrealistic extreme, because that extreme better separates the desired quality from the undesirable average (S-). This indicates that the learning mechanism involves abstracting the essential features of the S+ and maximizing them, leading to a preference for a stimulus that is “more S+” than S+ itself.

This aesthetic Peak Shift accounts for why caricature is often instantly recognizable and why abstract art can sometimes evoke stronger emotional responses than realistic depictions. A caricature exaggerates the distinguishing features of a face, pushing the visual information past the normal range to a point that maximizes the contrast and distinctiveness, thereby increasing recognition and attention. In essence, the psychological system prefers the stimulus that provides the clearest, most unambiguous signal of the desired attribute, even if that signal is biologically or structurally unnatural. The power of the exaggerated stimulus lies in its ability to avoid the ambiguity inherent in the average or typical stimuli, which serve as the implicit S-.

Significance and Applications in Psychological Science

The phenomenon of Peak Shift holds profound significance for psychological science, fundamentally altering how researchers conceptualize the nature of stimulus control and learning. It serves as a powerful demonstration that complex organisms do not merely absorb information passively but actively process and compare stimuli, learning relationships rather than just absolute values. This relational view has implications across various sub-disciplines, including cognitive psychology, comparative psychology, and even neuroscience, where efforts are made to identify the neural correlates of excitatory and inhibitory gradients.

In applied settings, understanding Peak Shift is crucial for effective training and education. When designing teaching methodologies that require fine discrimination, trainers must be aware that the inclusion of non-reinforced examples (S-) will not only inhibit responses to S- but also shift the optimal response criteria for S+. If a child is taught to identify a specific shade of blue (S+) and is explicitly shown a slightly different shade of blue not to touch (S-), the child may respond most strongly to an even deeper, more distinct shade of blue (the Peak Shift) rather than the original instructional color. Recognizing this displacement allows educators to adjust training parameters or reinforce the precise target stimulus more intensely to counteract the shift.

Ultimately, Peak Shift is a robust and highly replicable finding that confirms the predictive power of gradient summation models, demonstrating how internal inhibitory processes actively contribute to observable behavior. It provides a unifying principle linking basic associative learning mechanisms—where the avoidance of an unwanted stimulus reshapes the preference for a desired stimulus—with complex aesthetic choices, confirming that the search for optimal distinctiveness and exaggerated features is a pervasive psychological tendency across different domains of experience. The complexity revealed by Peak Shift underscores that learned responses are emergent properties resulting from the dynamic interplay between excitation and inhibition operating along a continuous stimulus dimension.