Criticisms of the biological preparedness models and alternative explanations

empirical support, criticisms of the biological preparedness perspective have been formulated (see, e.g., Davey, 1995; Mallan et al., 2013; McNally, 1987; see also Åhs et al., 2018). In particular, the biological preparedness accounts have been argued to overly rely on evolutionary explanations, without considering the importance of ontogenetic and cultural factors in modulating preferential emotional learning (e.g., Mallan et al., 2013). In agreement with this view, several findings have cast doubt on the putative superiority of threat-relevant stimuli from phylogenetic origin relative to threat-relevant stimuli from ontogenetic origin. For instance, Hugdahl and Johnsen (1989) reported that enhanced resistance to extinction to pictures of guns pointed toward participants (i.e., a cultural threat) associated with a loud noise as the US was not statistically different from resistance to extinction to snake pictures directed toward participants (i.e., a biological threat) associated with an electric stimulation as the US.

Whereas Öhman and Mineka (2001) acknowledged that the fear module can also be activated by ontogenetic threat-related stimuli under certain circumstances (e.g., extensive training) in addition to its preferential activation by phylogenetic threat-relevant stimuli, they however clearly argued that such activations should not occur to threat-relevant stimuli from ontogenetic origin when they are presented under suboptimal, potentially unaware conditions. In contrast to this prediction, Flykt, Esteves, and Öhman (2007) showed an enhanced resistance to extinction to both unmasked and masked presentations of snake and gun pictures during extinction when they were directed toward participants, but not when the masked snake and gun pictures were directed away from them. Jointly, these studies seem to indicate a similar conditioning effect across phylogenetic and ontogenetic threat-relevant stimuli, which is at odds with the tenets of the preparedness and fear module theories. Similar results were also

documented with respect to attention processes. Several experiments (e.g., Blanchette, 2006;

Brosch & Sharma, 2005; Fox, Griggs, & Mouchlianitis, 2007) have found that both phylogenetic and ontogenetic threats attract attention more quickly than neutral stimuli in a comparable manner. A recent study further raised the possibility that modern threats might even be detected faster than ancient threats (Subra, Muller, Fourgassie, Chauvin, &

Alexopoulos, 2017). Taken together, these empirical data suggest that the key factor modulating Pavlovian aversive conditioning and attentional biases to threat-relevant stimuli may be threat-relevance rather than evolutionary history per se.

In a similar vein, another line of evidence inconsistent with the preparedness and fear module theories concerns Pavlovian aversive conditioning to social threat-relevant stimuli, which appears to be more malleable than Pavlovian aversive conditioning to animal threat-relevant stimuli (Mallan et al., 2013). Whereas learned threat to animal threat-threat-relevant stimuli has been shown to be impervious to instructed extinction (Hugdahl, 1978; Hugdahl & Öhman, 1977; Lipp & Edwards, 2002; Öhman, Erixon, et al., 1975; Soares & Öhman, 1993), learned threat to social threat-stimuli has been reported to swiftly extinguish following instructed extinction and electrode removal both for angry faces (Rowles et al., 2012) and outgroup faces (Mallan et al., 2009). These results suggests that Pavlovian aversive conditioning to social threat stimuli is susceptible to alterations by cognition, unlike Pavlovian aversive conditioning to animal threat stimuli (Mallan et al., 2013). It has therefore been proposed that preferential Pavlovian aversive conditioning to social threat-relevant stimuli may hinge on sociocultural factors, such as negative stereotypes or social norms, rather than purely biological factors, such as biological preparedness (Mallan et al., 2013; see also Olsson et al., 2005). Accordingly, these observations underline the notion that both phylogenetic and ontogenetic factors may play an important role in preferential emotional learning.

Further, a growing body of research has likewise questioned the hypothesized preferential processing of threat-relevant stimuli (Lipp, Kempich, Jee, & Arnold, 2014). For instance, Lipp et al. (2014) conducted an experiment implementing a differential Pavlovian aversive conditioning paradigm in which images of snakes and wallabies were presented supraliminally and suboptimally using a binocular masking procedure transiently blocking awareness thereof. Results showed that supraliminal and suboptimal presentations of both image classes induced reliable differential Pavlovian aversive conditioning, thereby reflecting no preferential emotional learning to evolutionarily threat-relevant stimuli compared with cute, nonthreatening stimuli (i.e., wallabies), even under conditions of highly degraded input. In that

regard, these findings are strongly inconsistent with the predictions made by the preparedness and fear module theories.

Based on these considerations, alternative theories to the biological preparedness models have been elaborated for explaining evolutionary threat-relevance phenomena (see McNally, 1987; Öhman & Mineka, 2001, for reviews). We address some of these alternative theoretical models in the following sections.

Selective sensitization

Several authors (e.g., J. A. Gray, 1987; Lovibond, Siddle, & Bond, 1993) have argued that the enhanced responding to evolutionarily threat-relevant stimuli induced by their association with aversive outcomes is more parsimoniously explained by selective sensitization, a nonassociative process, rather than by selective associations. According to this account, phylogenetic threat-relevant stimuli are biologically predisposed to elicit heightened fear reactions; however, such preexisting response tendencies require specific conditions to emerge, such as a state of arousal or threat. During Pavlovian aversive conditioning, the mere threat of electric stimulation would hence be sufficient to selectively sensitize and boost responding to threat-relevant stimuli to a larger extent than to threat-irrelevant stimuli. In this context, Lovibond et al. (1993) suggested that the effects of enhanced resistance to extinction to evolutionarily threat-relevant stimuli originate from increased responding to threat-relevant stimuli from evolutionary origin due to selective sensitization and the fact that such heightened responding is maintained to the threat-relevant CS+ during acquisition through its systematic pairing with the aversive US. Inversely, sensitized responding to the threat-relevant CS- is reduced during acquisition as the threat-relevant CS- acquires inhibitory properties through the prediction of the US absence, thus amplifying the CS+/CS- differentiation to threat-relevant stimuli prior to the extinction phase. On the other hand, responding to threat-irrelevant stimuli is not sensitized; the CS+/CS- differentiation is consequently lower than to threat-relevant stimuli before extinction, which may in turn result in faster extinction comparatively (Lovibond et al., 1993).

Despite the fact that selective sensitization can evidently occur during Pavlovian conditioning (e.g., Lipp et al., 2015; Öhman, Eriksson, et al., 1975), it has, however, been suggested to be a relatively short-lived phenomenon (e.g., Lipp et al., 2015). Thus, selective sensitization has been argued to be insufficient to explain the long-lasting and lingering effects

typically observed in Pavlovian conditioning studies using threat-relevant stimuli in humans and non-human primates (Öhman & Mineka, 2001).

Expectancy bias model

Davey (1992, 1995) proposed an alternative model that starkly opposes to the biological preparedness perspective as well as the fear module encapsulation hypothesis. According to this alternative model, preferential Pavlovian aversive conditioning to threat-relevant stimuli relies on cognitive biases rather than biological preparedness. Specifically, Davey holds that preferential associations between threat-relevant stimuli and aversive events arise from an expectancy bias, which consists of heightened expectation of aversive outcomes following threat-relevant stimuli. This expectancy bias would be notably a key determinant of Pavlovian CRs (Davey, 1992). In line with this view, Davey (1992) showed that human participants had enhanced a priori expectancies that aversive events would follow threat-relevant stimuli compared with threat-irrelevant stimuli, even when they were instructed that no aversive stimulus would be delivered. Such differential a priori expectancies between threat-relevant and threat-irrelevant stimuli have been shown to extinguish with continued nonreinforcement and to be reinstated by CS-US pairings or explicit threat of an aversive US (Davey, 1992).

Differential expectancies have furthermore been reported to be translated into differential SCRs through exposure to the actual US either before or during the experiment (Davey, 1992), thereby highlighting that enhanced US expectancy after threat-relevant CSs is strongly associated with enhanced SCR to these stimuli under certain conditions (see also Dawson, Schell, & Banis, 1986). Importantly, expectancy bias is hypothesized to be essentially determined by ontogenetic, cultural factors, such as the CS dangerousness, without however excluding the possibility that such bias may reflect the complex interplay of evolutionary and cultural influences (Davey, 1995). Accordingly, the expectancy bias model can flexibly accommodate preferential Pavlovian aversive conditioning to both phylogenetic and ontogenetic threat-relevant stimuli (Flykt et al., 2007; Hugdahl & Johnsen, 1989).

Nevertheless, whereas Davey’s (1992, 1995) expectancy bias model has potential in explaining a wide range of behavioral data generated in the context of Pavlovian aversive conditioning to threat-relevant stimuli in humans, Öhman and Mineka (2001) raised several problems with this model. In particular, it is inconsistent with findings that have demonstrated persistent differential SCRs to evolutionarily threat-relevant stimuli, but not to threat-irrelevant stimuli, even though US expectancies to both of these stimulus categories were already extinguished (Schell, Dawson, & Marinkovic, 1991), thus suggesting a dissociation between

expectancy bias and SCRs (Öhman & Mineka, 2001). Additionally, the expectancy bias model has difficulty in explaining the results from observational conditioning in monkeys showing that lab-reared monkeys with no ontogenetic experience with snakes readily developed fear reactions to these stimuli in a selective manner (M. Cook & Mineka, 1989, 1990; Mineka &

Öhman, 2002). For these reasons, it has been argued that expectancies are not sufficient to account for the body of evidence supporting the preparedness model, but may conversely be consequences rather than causes of fear responding (see Mineka & Öhman, 2002; Öhman &

Mineka, 2001).

Conditioned stimulus salience

Given that physical properties of the CS amplifying its salience can enhance its conditionability (e.g., Mackintosh, 1975; Rescorla & Wagner, 1972), CS salience has been suggested as a putative alternative mechanism to biological preparedness to explain the preferential Pavlovian aversive conditioning to threat-relevant stimuli (McNally, 1987).

According to this view, threat-relevant stimuli are preferentially conditioned to threat not because of their threat-relevance, but due to their high salience.

Although more salient or intense stimuli – in the sense of physical or perceptual salience (see Footnote 1 here above) – have been reported to be more readily conditioned than less salient stimuli (e.g., Mackintosh, 1975; Pearce & Hall, 1980; Rescorla, 1988a; Rescorla &

Wagner, 1972), it has been demonstrated that neutral stimuli that are highly perceptually salient (i.e., with a high visual complexity) do not induce enhanced resistance to extinction relative to neutral stimuli with a lower perceptual salience (i.e., low visual complexity; Öhman et al., 1976, Experiment 2). These findings thereby indicate that physical salience alone does not provide a satisfactory explanation for the effects of preferential Pavlovian aversive conditioning observed with threat-relevant stimuli (McNally, 1987; Öhman & Mineka, 2001).

As mentioned previously (see chapter 2.3), the physical salience hypothesis also appears inconsistent with classical models of Pavlovian conditioning (Mackintosh, 1975; Pearce &

Hall, 1980; Rescorla & Wagner, 1972), which predict that the CR to more salient stimuli should extinguish more rapidly than the CR to less salient stimuli (see, e.g., Siddle et al., 1988).

Nonetheless, it is important to note that salience can be equally conceptualized as not being limited only to the stimulus’ physical characteristics but also encompassing the stimulus’

relative importance to motivational contingencies that relate to the organism’s needs and goals (Cunningham & Brosch, 2012; Öhman & Mineka, 2001; Rescorla, 1988a). From this

perspective, it could therefore be suggested that threat-relevant stimuli benefit from enhanced Pavlovian aversive conditioning because of their high motivational salience. To the best of our knowledge, such motivational salience hypothesis has not been investigated to date.

Interestingly, a motivational salience account is however closely related to the alternative model that we propose hereafter, as it likewise suggests that preferential Pavlovian aversive conditioning stems from the interaction between the stimulus at stake and the organism’s motivational state.

2.3.4. Relevance detection as a key mechanism underlying preferential emotional