- 234 Downloads
Biological preparedness is a broad explanation for why some associations are learned more easily than others, invoking the evolutionary history of the animal.
Animals appear to learn about some stimuli in the world better than others, and a number of studies have shown that some sets of stimuli seem to be more easily associated than others. Some explanations exist for why this might occur, including ideas of belongingness and preparedness. Biological preparedness was first introduced as a term by Martin Seligman in his seminal paper “On the generality of the laws of learning” (Seligman 1970). As with any trait, learning and behavior in any species are the result of evolutionary processes. Broadly speaking, learning enables an animal to use past experience to inform future behavior. Learning allows an individual to better track a changing environment, and this plasticity forms the foundation of flexible behavior. While the neurological mechanisms underlying learning are highly evolutionarily conserved and shared among animals, we see large differences in what animals can learn about, and what associations animals can make more easily than others. Seligman explained these differences as a product of the biology of the animal and of its evolutionary history: An individual of a species is born more prepared to learn about some associations than others because those associations were more useful for the ancestors of that individual. This makes intuitive sense when one considers that animals are bombarded with stimuli and potential associations from birth. If all stimuli were held equal, the extended trial and error learning required to differentiate among the potential stimuli and outcomes would likely result in incredibly high mortality in young animals. And, indeed, there are many examples of spectacularly fast learning in the natural world. For instance, butterflies can learn associations of where to lay eggs after single experiences (e.g., Snell-Rood and Steck 2015), and cues of predation or danger are very quickly learned from conspecifics (e.g., Cook and Mineka 1990; Ferrari et al. 2012).
History: From Behaviorism to Preparedness, Constraints, and Adaptive Specializations
Early in his writings on learning, Thorndike (1911) speculated that the neural connections present within an animal may constrain what it can learn. Twenty years later, Thorndike (1932) wrote that the elements within learning that “belonged” together are more easily associated than elements lacking such a relationship. He referred to this as belongingness. Thus, how well something is learned is not determined by contiguity alone. In Thorndike’s view, an animal that is thirsty would more readily learn an association that contained an effect of water because such an effect belongs to the needs of the organism. An effect that “belonged” less would not elicit as strong a reaction. In this way, animals would show better learning to more appropriate pairings of stimuli and effects than to less appropriate pairings. Thorndike’s view of belongingness emphasizes the particular states of the individual, but how should differences in species be interpreted?
The approach of behaviorism saw the testing of many animal subjects under tightly controlled and replicated conditions. It was generally hypothesized that learning was governed by a series of common laws, and detailed study of model species could then apply broadly across all animals. During the 1960’s and beyond, a larger variety of species and tasks were being used in studies of learning. This led to an increasing number of potential exceptions to the generally regarded laws of learning, including two frequently cited studies which provide good examples of selectively learned information. In a study with dogs, Dobrzecka et al. (1966) compared qualitative and directional auditory cues in differentiation learning instrumental task. In a right leg-left leg differentiation task, dogs almost exclusively learned using the directional cues with the qualitative cues not being learned even after more than 1000 trials. Complicating the conclusions further, they found that in a go-no go task, dogs acted with the quality of the cue 80% of the time, but acted with the direction of the cue 20% of the time. In this study, not only are certain stimuli and responses learned better than others, but the effect is task-specific. During the same year, Garcia and Koelling (1966) showed that rats select the cues used as stimuli dependent on the nature of the reinforcement. Rats were given water that differed in taste and in audiovisual characteristics: “bright-noisy” water and “tasty” water. Rats drinking these types of water were later made ill using X-rays and lithium chloride. Garcia and Koelling found that avoidance reactions developed based on the taste cues but not the audiovisual cues. However, when paired with shocks, rats learned the audiovisual cues better than the taste cues. Again, different cue types are being learned in different tasks. Both papers were thought to pose challenges to the assumed generalities of the laws of learning and sparked a debate on the possibilities and potential types of biological constraints on the general process of learning (e.g., Seligman 1970; Shettleworth 1972; Logue 1979; Damianopoulos 1989).
A major idea proposed to explain the phenomena found by Dobrzecka, Garcia, and others was that of biological preparedness. Seligman (1970) argued that animals are born biologically prepared to learn certain associations, meaning that there is no equivalence of associability across all learning and all animals. Animals can also be contraprepared to learn. Contraprepared learning describes the difficulties in conditioning animals with certain tasks, such as with having a cat lick itself to escape a box. Seligman argued that animals show a continuum of preparedness, where learning about different events could be prepared, unprepared, or contraprepared. Examples from classical and instrumental conditioning, and discrimination and avoidance learning were discussed, with preparedness argued to be relevant to each of these types of learning. Whether prepared learning is qualitatively different from general learning was left an open question by Seligman due to the lack of experiments at the time. Additional questions proposed by Seligman were whether prepared learning covaried with additional cognitive or physiological mechanisms.
Shettleworth (1972) proposed a more general idea of biological constraints on learning in a detailed review of such constraints in every step of the learning process. Biological constraints on learning are essentially species-specific reactions to certain learning situations which are overlaid upon the factors operating in traditional learning theory. In other words, associative learning is the same across all species and situations with the exception of these biological constraints. She criticized Seligman’s notion of preparedness as being focused on associative learning as the cause for observed differences in associative learning that also could be due to nonassociative processes. Shettleworth also discussed the importance of learning in the wild and described how psychological ideas of learning could be applied to behaviors such as movement patterns and orientation. Shettleworth suggested that in studying learning, the behavior of animals in the lab should be related back to their behavior in the wild. Robert Bolles wrote about specific examples of this, proposing that some responses are likely constrained by Species-Specific Defense Reactions (e.g., Bolles 1970). He used example of flight reactions in rats when presented with aversive stimuli such as shocks.
In the following years, ideas of biological constraints became more specific, and were spoken of as adaptive specializations (e.g., Rozin and Kalat 1971; Shettleworth 1994). The notion of adaptive specializations specifically addresses the covarying cognitive and physiological mechanisms mentioned by Seligman. The use of the term adaptive specializations implies that such differences are the result of natural selection. Within the specific context of memory, Sherry and Schacter (1987) described the evolution of specific memory and learning systems as a result of functional incompatibilities between different tasks. Evolved modules would be well-suited to performing different tasks, such as food storing and retrieval behaviors in birds (e.g., Shettleworth 1990), and these adaptive specializations could then explain examples of preparedness described in learning. The notion of adaptive specializations is not without critics, but certainly every exception to general processes of learning does not require an explanation rooted in adaptive specializations (e.g., Krause 2015). Constraints do exist, and arguments using evolutionary adaptive logic can and should be empirically tested.
Prominent Examples of Preparedness: Taste Aversion, Fear, and Phobias
The biological importance of taste aversion is straightforward to invoke as forming fast associations to enable avoiding potential toxicity in food is undoubtedly adaptive. An indeed, a taste aversion was invoked frequently as an early example of preparedness (e.g., Rozin and Kalat 1971; Logue 1979).
This logic is especially true for generalist foragers such as rats, with a well-described mechanism for social learning of food odors. The textbook example of preparedness is the work by Garcia and Koelling (1966). Among other results, they found that rats quickly formed associations between taste and nausea, but were unable to form associations between light and sound, and nausea. This work was the start of a burgeoning literature on biologically prepared taste aversion. Recent work on prepared learning and taste aversion in an ecological context is looking at how predators learn to avoid aposematically colored and toxic prey. Aposematic coloration, or warning coloration, poses an evolutionary question because the potential prey item is conspicuous to predators and every predator must first learn that association between coloration and toxin (Skelhorn et al. 2016). First, there is prepared learning for bitter tastes and toxicity, resulting from an evolved response to bitterness reliably predicting toxic effects (e.g., Skelhorn and Rowe 2010). Additionally, the conspicuousness of the prey enables an easier detection by the predator, is easier to learn, and less likely to be forgotten (e.g., Alatalo and Mappes 1996; Prudic et al. 2007; Roper and Redstone 1987). Thus, within one ecological paradigm, multiple levels of preparedness can be predicted and observed.
A second area with a deep literature on preparedness with fear and phobias, primarily in humans and other primates, is fear conditioning, which like avoiding toxic food, is easily tied logically to ecological relevance. Animal should more quickly form associations among stimuli that can result in their harm or death. The most famous of these examples is the work of Cook and Mineka (1990), who demonstrated selective associations at play in the observational conditioning of fear in rhesus monkeys. In their study, monkeys selectively associated snakes with fear and did not selectively associate more biologically neutral stimuli, such as flowers, with fear. Additional studies of fear conditioning generally confirmed these results across a range of ecologically relevant and neutral stimuli (e.g., Ohman and Mineka 2001). This type of fear conditioning has also been proposed to be evolutionarily tied to the development of phobias. Phobias in humans seem to develop around specific objects, animals, or situations and are not equally distributed among other types of stimuli; thus, it is widely thought that phobias may represent a particular case of preparedness (e.g., Seligman 1971; Ohman and Mineka 2001).
Mechanisms for Preparedness
A specific mechanism proposed for some examples of preparedness is that of selective associations. Selective associations require an interaction between the stimulus and reinforcement that results in faster acquisition of a CR with one CS than with another, but an opposite effect with a different reinforcement (LoLordo 1979). Additionally, the differences in learning must be due to this interaction and not to any differences to any particular properties of the stimulus used, such as its salience or intensity. This quite specific definition requires a series of controls to ensure that the observed differences are indeed due to selective associations, and LoLordo reviews possible methodologies to use for testing for a selective association. A straightforward approach is provided by Linwick et al. (1981), and they emphasized using correctly chosen controls and using a bidirectional design such that excitatory and inhibitory effects upon the dependent measure could be seen, without any floor or ceiling effects that might mask important differences. It is important to note that selective associations are not the only form of preparedness described in animals, as evidenced by the many other mechanisms that can result in prepared or contraprepared learning. Selective nonassociative processes may be adequate to produce the biologically appropriate, or prepared, response to the situation.
Given that natural selection acts upon fitness rather than mechanism, the output of the learning and not the process may be what should be considered from an evolutionary point of view. Seligman predicted that physiological and cognitive aspects would vary along with the continuum of preparedness for learning specific associations in animals. Evolutionary processes can prepare animals to learn some associations more easily than others through acting on multiple cognitive systems, from perception to motivation to learning and decision making. In a broader biological view of learning, we know that this must be the case. Harkening back to Shettleworth’s list of biological constraints, an additional 45 years of work on the neurobiology and sensory biology of cognition have finally enabled a more full con ception of mechanisms through which preparedness happens. For instance, there is currently a strong emphasis on the role of sensory bias in the evolution of animal behavior (e.g., Ryan and Cummings 2013; ten Cate and Rowe 2007), and a myriad of ways that such a bias might be accomplished across perceptual systems (Stevens 2013). Once perception of a stimulus is possible, both selective attention and sampling behavior, the tendency to encounter stimuli and the experiences which lead to forming associations can affect what is learned (e.g., Knudsen 2007; Snell-Rood and Steck 2015). Within nonassociative learning, we can predict that biological prepared learning may be enhanced through greater sensitization and more difficult to habituate to. Within classic associative learning, two relevant paradigms address how biological preparedness may enable certain associations to become learned over others. The first is that when multiple stimuli are present as the animal learns, the more salient stimulus can overshadow potential learning of another stimulus. The second is that a prior exposure to a salient stimulus can then block the learning of any future presentations of a different stimulus. Both overshadowing and blocking occur frequently in studies, and are a predicted consequence of preparedness (e.g., LoLordo 1979). Prepared learning should also be resistant to extinction: being persistent in performing a learned response in the absence of the original reinforcement (like a reward or a shock). We can also predict that biologically prepared learning should be forgotten less quickly. Work in both psychology and behavioral ecology predict that more important information should be remembered for a longer period of time; forgetting will be adaptive for certain stimuli and experiences (e.g., Dunlap et al. 2009; Ferrari et al. 2012; Kraemer and Golding 1997).
The majority of work to date on preparedness in learning has focused on post hoc explanations of observations or of patterns uncovered in behavioral experiments. There are exceptions to this. For instance, Domjan et al. (2004) hypothesized differences between arbitrary and ecologically relevant stimuli in sexual conditioning contexts. While predictions of preparedness in terms of fear conditioning and phobias have been made, empirical results can be mixed. A primary problem has been the lack of predictive models. The first general theoretical model on how preparedness can evolve focuses on the reliabilities of stimuli that can be learned about (Dunlap and Stephens 2014). Building upon earlier models of the evolution of learning broadly (e.g., Dunlap and Stephens 2009; Stephens 1991), reliability is set as a contingent probability of a given cue accurately predicting an outcome. Because statistical properties of change are both measureable and empirically tractable, this allows for direct tests of when prepared learning will evolve. And, indeed, using the techniques of experimental evolution, it is possible to evolve prepared learning about some modalities of stimuli over others, directly testing this theory; Dunlap and Stephens showed exactly this with fruit flies, evolving flies that learned about color better than odor, and other flies that learned about odor better than color, matching the predictions of the model. By making predictions from reliabilities of stimulus-outcome contingencies, preparedness can also be more directly tested within single lifespans, in addition to across generations.
Notions of biological preparedness have a long history in the study of animal cognition. Preparedness has most often been invoked to explain quick learning in high-risk situations: predation, harm, potential poisoning. While it is easy to imagine the evolutionary implications of death, biological preparedness can equally apply toward appetitive or sexual stimuli. Although much past work has focused primarily on the role of selective associations in preparedness, evolution selects on outcomes and not mechanisms and biological preparedness in learning can be accomplished through a number of mechanisms, both associative and nonassociative. The evolution of biological preparedness can be predicted through an analysis of patterns of reliability of relevant stimuli, thus providing a guide for empirical work rather than a reliance on post hoc explanations.
- LoLordo, V. M. (1979). Selective associations. In A. Dickinson & R. A. Boakes (Eds.), Mechanisms of learning and motivation: a memorial volume to Jerzy Konorski (pp. 367–398). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc..Google Scholar
- Rozin, P., & Kalat, J. W. (1971). Specific hungers and poison avoidance as adaptive specializations of learning. Psychological Review, 78, 459–486.Google Scholar
- Sherry, D. F., & Schacter, D. L. (1987). The evolution of multiple memory systems. Psychological Review, 94, 439–454.Google Scholar
- Shettleworth, S. J. (1972). Constraints on learning. In D. S. Lehrman et al. (Eds.), Advances in the Study of Behavior, 4, 1–68. Academic Press.Google Scholar
- Shettleworth, S. J. (1990). Spatial memory in food-storing birds. Philosophical Transactions of the Royal Society of London B, 329, 14–151.Google Scholar
- Shettleworth, S. J. (1994). Biological approaches to learning. In N. J. Mackintosh (Ed.), Animal Learning and Cognition (pp. 185–219). San Diego: Academic Press.Google Scholar