Factors influencing Pavlovian conditioning

Whereas a central aim of this thesis is to elucidate the mechanisms underlying why some stimuli are preferentially learned above others during Pavlovian conditioning, investigation of Pavlovian conditioning in animals and in humans has been mainly guided by the aim of uncovering the general laws of learning that apply across different kinds of stimuli, rather than highlighting differences between classes of stimuli. This line of research has notably identified a number of key variables that exert an influence on Pavlovian learning. Before addressing in more depth the theoretical models advanced to explain the occurrence of preferential Pavlovian conditioning in response to certain stimuli (see chapter 2.3), we briefly detail these key factors below.

Stimulus novelty and intensity

The CS and the US have been demonstrated to be more rapidly and effectively learned when they are novel at the beginning of conditioning (Bouton, 2007; Nasser & Delamater, 2016). Preexposure to the CS before Pavlovian conditioning can indeed interfere with learning by reducing the rapidity with which organisms learn about the CS-US association when they are subsequently paired, an effect called latent inhibition (e.g., Lubow, 1973; Vaitl & Lipp, 1997). Similarly, preexposure to the US alone before conditioning can reduce its effectiveness as a US and delay subsequent Pavlovian learning (Bouton, 2007), this effect being referred to as US preexposure effect (Randich & LeLordo, 1979).

The US intensity also critically influences Pavlovian learning, with more intense USs producing faster and stronger Pavlovian conditioning (Bouton, 2007; Nasser & Delamater, 2016; Rescorla & Wagner, 1972). Likewise, the intensity or “salience”¹ of the CS plays an

1 It is important to note that the construct of salience here refers to the stimulus’ physical properties (e.g., brightness, contrast, loudness, color intensity; e.g., Kamin, 1965; Öhman & Mineka, 2001), as traditionally used in the Pavlovian conditioning literature, and is closely related to the notion of stimulus intensity (e.g., Pearce &

Hall, 1980). Although it has been acknoweldged that stimulus salience is not only confined to mere stimulus’

charachteristics but also relates to the stimulus’ importance to the organism’s motivational contingencies (see, e.g., Öhman & Mineka, 2001; Rescorla, 1988a), the role of motivational salience, as opposed to physical or

important role by affecting the rate of learning; with more intense or salient CSs being generally learned more rapidly (Kamin & Schaub, 1963; Mackintosh, 1975; Pearce & Hall, 1980;

Rescorla, 1988a; Rescorla & Wagner, 1972). As salient CSs are particularly attention-grabbing, this effect may reflect the influence of attentional processes on Pavlovian conditioning, whereby CSs that are better attended to become more readily conditioned (Mackintosh, 1975; Pearce & Hall, 1980). Another phenomenon highlighting the importance of the CS salience relates to overshadowing (e.g., Kalat & Rozin, 1970; Mackintosh, 1976;

Pavlov, 1927), which corresponds to the observation that weaker Pavlovian conditioning occurs to a CS when it is combined with a more salient CS during conditioning trials.

Number of CS-US pairings

The number of CS-US pairings is one of the most basic variables affecting Pavlovian learning. Whereas Pavlovian conditioning can occur through a single CS-US pairing in specific cases (e.g., Garcia & Koelling, 1966; Öhman, Eriksson, et al., 1975; Seligman, 1970, 1971), conditioned responding typically increases with the number of CS-US pairings, even when controlling for the overall amount of time of the conditioning procedure (Nasser & Delamater, 2016).

Temporal and spatial contiguity

The time that elapses between the CS and the US is also a crucial factor in Pavlovian conditioning. Usually, Pavlovian conditioning is the most effective when the CS precedes the US, thus acting as a signal thereof, the optimal time interval between the CS-onset and the US-onset varying across species and conditioning paradigms. Accordingly, the most efficient form of Pavlovian conditioning is forward delay conditioning (see Figure 2.2A), in which the CS is first presented and then coterminates with the US. Because of its high efficiency in producing Pavlovian conditioning, we notably implemented this procedure in all the empirical studies reported in the experimental part of this thesis (see chapter 3). Another form of forward conditioning is trace conditioning (see Figure 2.2B), where the CS begins and ends before the US is presented, the interval between the CS and the US being called the trace interval. Trace conditioning is in general less effective than delay conditioning and becomes less efficient the longer the trace interval. A third procedure is to present the CS and the US simultaneously in time, which is referred to as simultaneous conditioning (see Figure 2.2C). Simultaneous

perceptual salience, remains largely underinvestigated in Pavlovian conditioning. This distinction will be an important point for discussion later on.

conditioning generally causes weaker Pavlovian conditioning than forward conditioning procedures. A final procedure is backward conditioning (see Figure 2.2D), in which the CS is presented after the US and which is usually the least effective in producing Pavlovian conditioning, though it might sometimes result in conditioned inhibition, the CS becoming a signal for the absence of the US (Bouton, 2007).

Albeit much less explored than temporal contiguity, spatial contiguity has also been shown to influence Pavlovian learning (see Nasser & Delamater, 2016). More precisely, learning can be promoted when the CS and the US are contiguous in space (Nasser &

Delamater, 2016).

CS-US similarity

Although research on this specific factor is scant, CS-US similarity has been hypothesized to play an important role in the formation of an association between the CS and the US (e.g., Rescorla & Holland, 1976). In agreement with this view, Pavlovian conditioning

A. D

B. T

C. S

D. B

CS

US

CS

US

CS

US

CS

US TIME

Trace interval

Figure 2.2. Forms of Pavlovian conditioning implementing different time intervals between the conditioned stimulus (CS) and the unconditioned stimulus (US). Adapted from Bouton (2007).

is generally enhanced when the CS and the US are more than less similar (Nasser & Delamater, 2016).

US surprisingness or prediction error

The extent to which the US is surprising (i.e., because of a prediction error) is a fundamental determinant of Pavlovian learning. This idea originated from Kamin’s (1969) seminal work on blocking, based on which he suggested that Pavlovian conditioning only happens when the US is surprising. According to this view, blocking occurs because the CS previously associated with the US already predicts its occurrence, and the US is therefore no longer surprising (i.e., there is no prediction error) when a new CS is presented along with the CS, no learning occurring to the new CS as a result. The influence of US surprisingness has been furthermore supported by evidence showing that surprising USs are more effective reinforcers than expected USs (Rescorla & Wagner, 1972), thereby reflecting the key role of prediction error in learning (Niv & Schoenbaum, 2008; Rescorla & Wagner, 1972).

CS-US relevance or belongingness

The notion of CS-US relevance or belongingness refers to the fact that some combinations of CS and US are more readily conditioned (“belong” together) than others depending on their functional relation. Garcia and Koelling’s (1966) experiment on taste aversion learning provided the most compelling support for this notion. In this experiment, thirsty rats drank water that was paired with an audiovisual stimulus and a gustatory stimulus (bright-noisy-flavored water). Consumption of the water was paired with either nausea-inducing drug or footshock as the US in different groups of rats. Rats were subsequently tested several days later either with the flavored water in the absence of the audiovisual features, or with the bright-noisy water in the absence of the flavor. Results demonstrated that rats trained with the illness US avoided drinking the flavored water but consumed the bright-noisy water, whereas rats trained with the footshock US avoided consuming the bright-noisy water but drank the flavored water. This result hence indicated that specific CS-US associations can be learned preferentially. These selective associations provided evidence for the existence of biological constraints on learning (Garcia & Koelling, 1966; Seligman, 1970), and challenged the idea that all stimuli are equally associable and follow the same general laws of learning. Of importance, selective associations have also been demonstrated in human Pavlovian aversive conditioning in response to threat-relevant stimuli (e.g., Öhman et al., 1976; Öhman & Mineka, 2001; Hamm, Vaitl, & Lang, 1989; Hugdahl & Johnsen, 1989; see chapter 2.3).