Mechanisms of extinction in emotional regulation james Byron neLson
Associative-learning processes endow stimuli with the ability to elicit powerful emotions that both produce behaviors in their own right, and motivate behaviors that manage those emotions (see Nelson and Bouton, 2002). The relationships that establish associations between stimuli, however, are not static and the learning process must be able to adapt to changing contingencies. Just as conditioning is a way in which adaptation occurs through associative learning that allows important events to be expected, extinction is a way in which we learn that ordinarily expected events no longer occur. A person that has a traffic accident on the way to work will likely experience anxiety on that route in the future. Such anxiety may motivate him or her to take alternate routes to control the anxiety. In the laboratory, a tone might signal an imminent shock, eliciting fear in a rat. Given the opportunity to escape the signal and avoid the shock, the rat will undoubtedly do so. These initially neutral stimuli are conditioned stimuli (CSs) that acquire their ability to elicit and modulate emotions through their pairing with inherently significant stimuli (unconditioned stimuli or USs). With exposure to the travel route (in the absence of further mishap) anxiety should subside. With exposure to the tone alone, the rat’s fear will be reduced. In both situations, the CS now occurs without the US, which procedurally defines extinction. Theoretically, we can also say that a process of “extinction” took place. This chapter discusses processes that might be implied by the latter use of the term. The discussion will be devoted, in large part, to the observation that extinction leaves the original learning more or less intact, and does not involve unlearning. Thus, the discussion will omit theories that are important, but do not necessarily address the representational content of what could be learned during extinction or do not provide a mechanism through which initial learning can remain intact following extinction (e.g., Capaldi, 1967; Gallistel and Gibbon, 2000). Pavlov suggested that the representation of the CS undergoes modification during extinction such that it would be unable to activate the representation of the US. Support for this idea can be found in an experiment on “negative-patterning” in pigeons by Robbins (1990). Pigeons first learned a discrimination where two visual CSs were paired with a food US when presented alone (i.e., CS1−US, CS2− US) but not when presented together sequentially (i.e., CS1→CS2 – No US). Birds
first mastered the discrimination showing strong responding to either CS alone, but not when presented sequentially. Next the birds underwent extinction with CS1. The ability of CS1 and CS2 to jointly control non-responding when presented together was believed to be a function of the internal representations of these stimuli. Therefore, if the reduction in responding to CS1 during extinction was due to a change in the internal representation evoked by CS1, the reduction would paradoxically weaken CS1’s ability to control non-responding in the CS1→CS2 compound; exactly the result obtained. Extinction of CS1 produced a decrease in responding to CS1, and an increase in responding to the CS1→ CS2 compound. The CS−focused approach to extinction proposed by Pavlov links extinction to processes associated with habituation, opening a variety of additional theoretical avenues by which extinction might be better understood. Such a case is presented in detail by McSweeney and Swindle (2002). Other theories can be portrayed as involving “interference”. One idea represented today in the work of Wagner (1981), assumes that “excitatory” associations are learned between the CS and US during conditioning so that the CS can evoke an internal representation of the US. During extinction, an “inhibitory” association is formed that interferes with the ability of the excitatory association to elicit the US representation, leaving the CS functionally neutral. This idea makes many of the same predictions as the subtly distinct one put forth by Jerzy Konorski, where excitatory associations are formed during extinction between the CS and a representation of the absence of the US, or “No-US” representation. This No-US representation suppresses the activation of the US representation. In this theory, the interference is not between associations competing to activate a single representation, but between different activated representations. In both, the original learning remains intact during extinction upon which new learning is superimposed. Research on “conditioned inhibition” is relevant to this latter interpretation of extinction as conditioned inhibition is assumed to be fundamentally the same as that inhibition acquired during extinction. In a conditioned inhibition procedure, one stimulus signals a US (e.g., CS1−US) but not when combined with another stimulus (CS1CS2−No US). In these designs, the organism treats CS2 as if it predicts the absence of the US. CS2 will suppress the emotional reactions elicited both by CS1, and other CSs conditioned with the same US. There are examples where an extinguished stimulus also shares these properties (e.g., Calton, Mitchell, and Schachtman, 1996). Inhibitors appear to elicit a representation with its own emotional characteristics. Pigeons, for example, will actively avoid an inhibitor for food. Additionally, an inhibitor for shock seems to be motivationally similar to an excitor for food, and vice-versa (see Dickinson and Dearing, 1979). These observations are fundamental in appreciating that extinction is an active process with its own significance. Extinction elicits emotional states associated with the absence of the US, such as frustration when an expected favorable US does not occur, or relief when an expected aversive US is omitted. Extinction involves learning about the absence of the US; a representation with its own emotional properties which can motivate behavior.