Evolution and Human Behavior

Transcription

1 Evolution and Human Behavior 38 (2017) Contents lists available at ScienceDirect Evolution and Human Behavior journal homepage: Original Article Do infants associate spiders and snakes with fearful facial expressions? Stefanie Hoehl a,b,,sabinapauen b a Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany b Institute of Psychology, Heidelberg University, Heidelberg, Germany article info abstract Article history: Initial receipt 14 March 2015 Final revision received 1 December 2016 Keywords: Infancy Fear Social learning Preparedness ERP Do infants preferentially learn to fear stimuli that represent an ancestral danger? This question was addressed using event-related brain potentials in 9-month-old infants (N = 38). In Experiment 1, infants saw fearful and neutral faces gazing towards spiders and flowers. Then spiders and flowers were presented again without faces. Infants responded with increased attention (signaled by the Negative central, Nc component) to stimuli associated with fear. In particular, spiders that were gaze-cued with a fearful as compared to a neutral expression elicited an increased Nc response. In Experiment 2, targets were snakes and fish. Snakes elicited increased Nc amplitude compared to fish irrespective of emotion condition. Results speak to the evolution-based fearrelevance of spiders and snakes. Our findings provide partial support for social fear learning and preparedness theory (Experiment 1) and non-associative accounts of fear acquisition (Experiment 2). We conclude that both kinds of fear acquisition seem to play a role in early human development Elsevier Inc. All rights reserved. 1. Introduction Why are specific phobias often directed at ancestral threats like spiders and snakes even in regions of the world where these animals do not currently pose a deadly threat to humans (Fredrikson, Annas, Fischer, & Wik, 1996)? Here, we address this question from a developmental perspective by testing predictions from different theoretical accounts proposing fear acquisition as an evolutionary adaptation. To explain the development of specific phobias, several accounts have been presented in the literature. All of these accounts have in common that specific phobias are considered side effects of adaptive mechanisms of fear acquisition that have evolved in primate phylogeny. At the core of these theories lies the assumption that evolution has provided us with the ability to avoid potential dangers through fear without having to experience direct fear conditioning, i.e. without first having to put ourselves in harm's way. Three theories we discuss here are social learning, preparedness theory, and non-associative accounts of fear acquisition. These accounts differ in terms of the assumed content dependency of the underlying mechanisms. Whereas social learning as a general mechanism can be considered content independent, the other two theories posit that fear responses are more likely to occur for specific contents representing ancestral threats such as spiders. According to preparedness theory these content specific responses depend on input in the form of vicarious or direct learning experiences. According to Corresponding author at: Max Planck Institute for Human Cognitive and Brain Sciences, Stephanstr. 1a, Leipzig, Germany. Tel.: ; fax: address: hoehl@cbs.mpg.de (S. Hoehl). non-associative accounts, in contrast, content specific fear responses towards ancestral threats are deployed without requiring learning Social learning In daily life, the sight of a spider or a snake is rarely accompanied by pain and injuries due to bites in most regions of the Western world. Therefore, it has been suggested that observational learning rather than direct conditioning may nowadays be more relevant for the development of specific phobias(mineka, Davidson, Cook, & Keir, 1984; Olsson & Phelps, 2007; Rachman, 1977). In observational learning the fearful emotional expression of a conspecific is supposed to function as the unconditioned aversive stimulus which is associated with the conditioned stimulus leading to the learning of a fear-response (Olsson & Phelps, 2007). This process is not content dependent with regard to the conditioned stimulus, i.e. in principle any stimulus can be associated with fear. Infants are sensitive to emotional facial expressions from early on. By 7 months of age infants pay increased attention to fearful faces compared with happy or neutral faces (Nelson & de Haan, 1996; Peltola, Leppänen, Maki, & Hietanen, 2009), especially when fear faces look towards a potential threat in the environment (Hoehl, Palumbo, Heinisch, & Striano, 2008; Hoehl & Striano, 2010). When encountering a novel object, infants by 9 to 12 months of age search for emotional signals from adults (i.e., they show social referencing) and they adjust their behavior towards the object in accordance with the adults' emotional facial and vocal expressions (Hertenstein & Campos, 2004; Hornik, Risenhoover, & Gunnar, 1987; Moses, Baldwin, Rosicky, & Tidball, 2001; Mumme & Fernald, 2003) / 2016 Elsevier Inc. All rights reserved.

2 S. Hoehl, S. Pauen / Evolution and Human Behavior 38 (2017) Thus, infants are guided in their behavior by the emotional reactions and instructions they receive from more experienced others. Social learning can contribute to the emergence of pathological fear when infants and children regularly observe their caregivers' fear reactions in certain situations. For instance, infants of mothers with social phobia increasingly avoid strangers over time after observing their mothers' anxious social behavior (Murray et al., 2008). Infants' and children's strong propensity for social learning enables fear learning without direct fear conditioning, thus allowing for safer learning. However, for children of parents with phobias this means that they are at a high risk to develop an anxiety disorder themselves (Ginsburg, 2009). Social learning ensures that humans are able to learn from others' reactions early on in development. This alone does not explain why contents of specific phobias are more often ancestral threats than modern threats, such as electric outlets and guns. According to accounts focusing on social fear learning, infants should readily associate a fear expression with any kind of stimulus. Below, we therefore discuss two theories accounting for selectiveness in fear acquisition by proposing evolved content specific mechanisms Preparedness theory Seligman suggested in the early 1970s that humans have evolved a preparedness to associate fear with animals and situations that have posed a threat in human phylogeny (Mineka & Öhman, 2002; Seligman, 1971). According to preparedness theory, phobias are more likely to be acquired for animals and situations that were dangerous to our ancestors (e.g., spiders, snakes) than to objects that are dangerous nowadays but not in earlier phylogeny (e.g., guns; Mineka & Öhman, 2002). Thereisconsiderableevidenceforcontentspecificfearacquisition.In fear conditioning experiments with adults, certain stimuli representing an ancestral danger to humans, such as snakes and spiders, are more readily associated with unconditioned aversive stimuli (e.g. mild electric shocks) than other stimuli, such as flowers (see Öhman & Mineka, 2001, for a review). Furthermore, associations of ancestral threats with aversive sensations and fear seem to be more robust and less prone to extinction than associations with non-threatening stimuli (Cook, Hodes, & Lang, 1986; Öhman & Mineka, 2001). Relevant research on preparedness using social fear learning tasks has been conducted with non-human primates. Lab-reared rhesus monkeys do not normally show fear of snakes. However, they are able to quickly learn fear reactions from watching wild rhesus monkeys' reactions to snakes (Mineka et al., 1984). In a further experiment labreared monkeys saw conspecifics react fearfully to snakes and crocodiles or to flowers and toy rabbits using manipulated video recordings (Cook & Mineka, 1989). Only when monkeys saw fear reactions of their conspecifics to snakes and crocodiles they acquired fear of these stimuli. Together, these results support the notion that primates can learn fear of certain stimuli via observation, thus indicating the relevance of social information in the process of developing fear responses. In addition, primates seem to be prepared to learn fear of animals that represent ancestral threats speaking for content specific mechanisms underlying fear learning. In this view, infants should be prone to associate fear expressions especially with stimuli representing an ancestral threat Non-associative fear acquisition accounts Alternatively, proponents of non-associative fear acquisition accounts suggested that fears of certain animals and situations do not require direct or vicarious fear learning experiences (Menzies & Clarke, 1995; Poulton & Menzies, 2002). In this view, humans and other primates initially tend to fear certain stimuli and then habituate when experiencing these stimuli as harmless. Thus, the fear system is supposed to be highly content specific without requiring initial input. Support for non-associative fear acquisition comes from retrospective reports of phobic individuals that rarely contain fear conditioning experiences as well as prospective studies showing that, for instance, individuals who later report fear of heights did not experience more falls resulting in injuries during childhood (Poulton & Menzies, 2002). Taking all empirical evidence and theoretical accounts together, there is currently no definite answer to the question whether humans and other primates are especially prone to building fear associations with stimuli representing an ancestral danger or whether they initially respond with fear to these stimuli without requiring previous learning experiences. It is also possible that all of the discussed accounts are valid: Humans may initially be fearful of stimuli that have posed a threat in their phylogeny and then normally habituate through safe exposure, but they may still be more likely to associate these stimuli with fear later on (i.e., to re-learn the fear) through direct conditioning and/or social learning. A valuable approach to test these accounts is to examine infants' reactions to different kinds of stimuli associated with fear. In adults and older children the cultural context and prior encounters with spiders and snakes, even when not consciously remembered, may affect their responses. Testing infants, in contrast, allows us to trace back the origins of selective fear to its very beginnings in human ontogeny Infants' responses to ancestral threats Several studies have tested attention biases for stimuli representing an ancestral danger in infants (LoBue & DeLoache, 2010; Rakison & Derringer, 2008). Infants in these studies, similar to older children and adults (LoBue & DeLoache, 2008), were able to detect snakes and spiders and responded more quickly to these stimuli compared to control pictures depicting other animals. Mechanisms enabling quick detection of spiders and snakes may constitute an adaptation to ancestral environments in which venomous bites from these animals were a real threat to survival. Adults are able to detect a single briefly presented task-irrelevant spider in an inattentional blindness task (but they less often detect modern threats or houseflies) suggesting that the human visual system retains biases to reflectively direct attention towards this ancestral threat (New & German, 2015). These findings are highly interesting but they cannot answer the question whether fear reactions to snakes and spiders require learning because no fear reactions were recorded and no fear associations were induced. The attention biases for ancestral threat stimuli reported in infants and adults may, however, support selective social fear learning. To our knowledge, only two studies have tested the selectivity of fear associations for stimuli that constitute an ancestral danger in infants (Deloache & LoBue, 2009; Rakison, 2009). In the study by DeLoache and LoBue (2009) 8- and 16-month-old infants looked longer at movies of snakes but not other exotic animals when listening to a frightened human voice than when listening to a happy voice. In the study by Rakison (2009) 11-month-olds were habituated to a schematic fearful face displayed next to a spider or a snake. At test, female but not male infants looked longer at a novel picture of a spider or a snake paired with a positive schematic face compared with a mushroom or a flower paired with a positive schematic face. No corresponding effect was found if infants were initially habituated to a schematic fearful face displayed next to a mushroom or a flower. These studies support the idea that social information influences infants' behavioral responses especially to stimuli representing an ancestral danger, but corresponding evidence is still sparse and seems to be limited to certain kinds of stimuli (e.g., moving but not static snakes, DeLoache & LoBue, 2009) and populations (e.g., girls but not boys, Rakison, 2009). The goal of the current study is to further examine the viability of the social learning account and preparedness theory in early human development, using brain measures rather than looking times, and photographs of natural faces instead of schematic faces (Rakison, 2009) or tone of voice (Deloache & LoBue, 2009).

3 406 S. Hoehl, S. Pauen / Evolution and Human Behavior 38 (2017) Event-related potentials (ERPs) are transient changes in brain activity recorded from the scalp that are time-locked to external or internal events, e.g. pictures. ERPs reflect synchronized activity of cortical neurons and can inform us about perceptual and cognitive processes in response to different types of stimuli (de Haan, 2007). The use of ERPs allows for the detection of differences in the amount of attention allocated towards a visual stimulus even in the absence of differences in overt looking behavior (Reynolds & Richards, 2005). It is therefore sometimes a more sensitive measure than looking times (de Haan & Nelson, 1997; Hoehl, Reid, Mooney, & Striano, 2008; Snyder, 2007). ERPs allow for testing whether limitations of previously reported effects to certain groups (i.e., female infants, Rakison, 2009) andtypes ofstimuli (i.e., only moving but not static stimuli, Deloache & LoBue, 2009) are also found when applying a more sensitive measure of attention than in previous research. In Experiment 1, 9-month-olds were presented with fearful and neutral faces looking towards a spider or a flower. Then the spider or flower was presented again alone and infants' brain responses were assessed. In Experiment 2, 9-month-olds were tested with the same paradigm featuring snakes and fish. We decided to run two experiments rather than one in order to check for the generalizability of findings across stimulus categories. Amplitude of the negative central (Nc) component of the ERP was used as a measure of infants' attention in this study. Amplitude of this component correlates with other measures of infant attention (Richards, 2003), and it is sensitive to the emotional content of images as it is increased in amplitude for fearful faces compared with happy faces in 7-month-olds (Nelson & de Haan, 1996; Peltola et al., 2009). Based on preparedness theory,we predict to find an enhanced Nc amplitude, indicating increased attention especially for spiders and snakes that were associated with a fearful vs. neutral facial expression, whereas no such effect is expected for flowers and fish. A non-associative fear acquisition account, in contrast, would predict generally increased attention for stimuli representing an ancestral danger, i.e., a greater amplitude of the Nc for spiders and snakes as compared to flowers and fish, respectively, regardless of the emotional expression of the face. Alternatively, infants may respond with increased Nc amplitude to stimuli associated with a fearful facial expression regardless of stimulus category as reported in previous studies with other age groups and different stimuli (e.g., Carver & Vaccaro, 2007; Hoehl & Striano, 2010; Hoehl, Wiese, & Striano, 2008). This would speak to infants' general propensity for social learning independent of biases to fear certain kinds of objects or animals. 2. Experiment Material and methods Infants took part in an ERP study with a 2 2 within-subject design. Independent variables were emotion condition (fearful vs. neutral) and stimulus category (spider vs. flower) Participants The study was conducted in a medium-sized German town with participants recruited from a primarily middle-class socio-economic background. Contact details of parents with young infants were provided by the local residents' registration office. Parents were contacted by mail and invited to join our database of families interested in participation in our studies. Parents of 8 9-month-olds registered in our database were then contacted by phone ahead of testing. They were informed about the particular study and a date for testing their infant was arranged if they were interested. All participating infants were born full term (37 41 weeks) and were in the normal range for birth weight. Eighteen infants were included in the sample (10 females, age range: 274 days to 298 days; average age: 285 days). Another 23 infants were tested but excluded from the sample because they failed to reach the minimum requirement of 10 artifact-free trials per condition for averaging. Infants that were excluded from the sample had seen a similar amount of trials in each condition, i.e., they did not drop out specifically because they avoided looking at spiders. Our criterion for inclusion and the resulting attrition rates are typical for infant ERP studies (DeBoer, Scott, & Nelson, 2007; Stets, Stahl, & Reid, 2012). All experiments were conducted with the understanding and informed consent of each participant's parent Stimuli Stimulus material consisted of portrait photographs of one male and one female actor whose eyes were directed either to the left or to the right in a horizontal plane. 1 Neutral and fearful pictures were taken from both actors. Their pupils and irises were moved from the middle to the left and right corner of the eyes using Adobe Photoshop CS2. A small picture of a spider or flower was displayed next to the face either to the left or right side, at the height of the pupils of the face, approximately 2 cm away from the eyes (see Fig. 1 for stimulus examples). The eyes were therefore effectively directed at the targets. Eight different exemplars of each target category (spiders, flowers) were used. Each flower was matched with one of the spiders for color and size using Adobe Photoshop CS2. The same targets were presented in both emotion conditions (fearful or neutral) equally often. Each of the different spiders and flowers was looked at with a fearful expression by one actor and with a neutral expression by the second actor. The same actor showed a fearful and a neutral facial expression equally often. No consistent information about the valence of individual target stimuli was provided because we were interested in the immediate effects of emotional expressions on infants' attention towards the respective objects. We did not intend to induce long-term learning effects which would have been ethically problematic. Stimuli showing a facial expression combined with a target were 25 cm long (from the outermost edge of the target to the actor's ear on the opposite side of the picture, visual angle of 15 ) and 19.5 cm high (visual angle of 12 ). Stimuli showing a single target in a central position were 5 x 5 cm of size (visual angle of 3 ) Procedure Infants sat on their parent's lap in a dimly lit room, at a viewing distance of 90 cm away from a 70 Hz 17-inch stimulus monitor. Experiment 1 consisted of one block with 160 trials (40 trials fearful-spider, 40 neutral-spider, 40 fearful-flower, 40 neutral-flower). Stimuli were presented using the software ERTS (BeriSoft Corporation, Germany). The four within-subject conditions were presented to the infant in a random order with the constraint that the same condition was not presented three times consecutively and that the number of presentations of each set of stimuli was balanced in every 32 trials presented. Each trial consisted of a central attractor object (displayed for 500 ms), a fearful or neutral face plus target stimulus (displayed for 1500 ms), a blank screen period with a randomly varying duration ( ms, on average 500 ms) and a target displayed at the centre of the screen (displayed for 1000 ms). ERPs were time-locked to the onset of the target presentation. Every trial was followed by a blank screen period, whose duration varied randomly between ms (on average 1000 ms). If the infant became fussy or uninterested in the stimuli, the experimenter gave the infant a short break. The session ended when the infant's attention could no longer be attracted to the screen. EEG was recorded continuously and the behavior of the infants was also video-recorded throughout the session. 1 Original pictures were taken from the MacBrain Face Stimulus Set. Development of the MacBrain Face Stimulus Set was overseen by Nim Tottenham and supported by the John D. and Catherine T. MacArthur Foundation Research Network on Early Experience and Brain Development (Tottenham et al., 2009).

4 S. Hoehl, S. Pauen / Evolution and Human Behavior 38 (2017) Fig. 1. Stimulus examples. Note that in the actual stimulus presentation a different male and a female were used EEG recording and analyses EEG was recorded with a 32 channels ActiCap system (Brain Products, Gilching, Germany) containing active electrodes based on Ag/ AgCl sensors, which were attached to an elastic cap and mounted according to scalp locations of the system. Data were amplified via a BrainAmp amplifier at a sampling rate of 250 Hz. The right mastoid electrode was used as online reference and an online filter was set at Hz. Horizontal and vertical electro-oculograms were recorded bipolarly. EEG data were re-referenced offline to the linked mastoids. A bandpass filter was set from Hz. The EEG recordings were segmented into epochs of 1200 ms including a (blank screen) baseline of 200 ms prior to stimulus-onset. A baseline correction was applied. Electrical artifacts caused by eye and body movements were rejected offline using automatic artifact detection methods of ERPLAB. Only trials were included in which amplitude of the analyzed channels did not exceed a voltage threshold of 200 μv within a 200 ms moving window. This window was moving in steps of 100 ms. Additionally, blinks were rejected in a semi-automatic procedure rejecting trials in which the EEG data exceeded a normalized crosscovariance threshold of.7 within a 400 ms wide time period, if sufficient EOG data were available. Based on the infant's video coded behavior, only trials were included when the infant saw the whole sequence of the trial and performed no obvious movements or blinks during the presentation of the target. Each infant contributed 10 to 28 trials in the fearful spider condition and 10 to 31 trials in the neutral spider condition. Furthermore, each infant contributed 10 to 28 trials in the fearful flower condition and 10 to 32 trials in the neutral flower condition. There was no difference in the average number of usable trials in each of the four conditions: an average of 14 trials was contributed per infant per condition in each of the four conditions Results Huynh-Feldt corrections were employed where applicable in all reported statistical tests and level of significance was set at p b Time-windows and included electrode sites were chosen based on visual inspection of the grand average and on the existing literature (Hoehl & Striano, 2010). Statistical analyses were carried out using IBM SPSS. In preliminary analyses we included sex of the infant as between-subject factor, because a previous study reported sex-specific effects in 11- month-olds' learning of fear-object associations (Rakison, 2009). No main effects or interactions with sex were found in either of the two experiments, so this factor will not be considered in the further analyses. ERPs for the spiders in the fearful and neutral condition are shown in Fig. 2. Fig. 3 showserpsfortheflowers in the fearful and neutral condition. An enhanced negativity is clearly visible in the fearful spider condition compared to the neutral spider condition spreading over frontocentral electrode sites where the Nc is typically measured (Reynolds & Richards, 2005; Richards, 2003). No such difference is observed in the flower condition. For statistical assessment mean amplitude between ms after stimulus onset was taken as the dependent variable in a Repeated Measures Analysis of Variance (ANOVA) in order to assess differences in Nc amplitude across conditions. In further analyses peak amplitude and latency to peak of the Nc were analyzed. Visual inspection of the waveforms suggested that there might also be an effect of emotion in the spider condition in an earlier time-window, i.e., a visible difference between the conditions occurred prior to the typical timewindow of the Nc (see Fig. 2). The positive deflection in this timewindow is usually referred to as the Positive before (Pb) component (Karrer & Ackles, 1987). Therefore, amplitude of the Pb was also analyzed between ms after stimulus onset. Within-subject factors for all Repeated Measures ANOVAs were emotion condition (fearful, neutral), stimulus category (spider, flower), and electrode site (i.e., position on the scalp according to the international system: F3, C3, FC1, FC2, FZ, CZ, F4, C4) Nc: ms A significant main effect of emotion condition was found for amplitude of the Nc, F(1,17) = 5.07, p = 0.038, partial η 2 =.23.Mean amplitude was greater for spiders and flowers in the fearful condition (mean = μv, standard error = 2.5) relative to spiders and flowers in the neutral condition (mean = 4.56 μv, standard error = 3.5), suggesting that 9-month-olds directed more attention towards stimuli cued by a fearful face compared to stimuli cued by a neutral face, irrespective of stimulus category (i.e., spiders or flowers). There was no significant main effect of stimulus category, p = 0.458, and no interaction of stimulus category by emotion condition, p = The same main effect of emotion condition was found when peak amplitude instead of mean amplitude between ms was used as dependent measure, F(1,17) = 4.53, p = 0.048, partial η 2 =.21.No other significant main effects or interactions involving emotion condition or stimulus category were found for mean amplitude or peak amplitude and no effects were found for latency to peak of the Nc, all psn 0.2. To test our a priori hypothesis that fearful expressions have a greater effect on infants' brain responses to spiders compared to flowers, we conducted two additional planned analyses with the factors emotion condition and electrode site for each stimulus category (spiders, flowers) separately. When considering only the trials showing spiders, we found a significant main effect of emotion condition, F(1,17) = 5.46, p = 0.032, partial η 2 =.24. Mean amplitude of the Nc was increased for spiders in the fearful condition (mean = μv, standard error = 2.9) compared to the neutral condition (mean = 1.52 μv, standard error = 5.3). The same main effect of emotion condition was found when peak amplitude instead of mean amplitude between ms was used as dependent measure, F(1,17) = 4.51, p = 0.049, partial η 2 =.21.Noothersignificant effects were found, neither for mean amplitude, nor for or peak amplitude, or latency to peak of the Nc, all psn 0.1. When considering only trials showing flowers, no main effect of emotion condition and no interaction involving emotion condition were found, all psn 0.2. Mean amplitude of the Nc was only slightly more negative for flowersinthefearfulcondition(mean= μv, standard error = 3.3) than in the neutral condition (mean = 7.6 μv, standard error = 4.1). Thus, the main effect of emotion condition that was found in the overall analysis was mainly driven by the spider condition.

5 408 S. Hoehl, S. Pauen / Evolution and Human Behavior 38 (2017) Fig. 2. ERPs time-locked to the onset of the spiders in Experiment 1. Infants showed a significantly increased Nc for spiders in the fearful condition (gray line) compared to the neutral condition (black line). Negative is plotted upwards Pb: ms Asignificant main effect of emotion condition was found for amplitude of the Pb, F(1,17) = 6.2, p =0.023,partialη 2 =.27. Mean amplitude was reduced for spiders and flowers in the fearful condition (mean = 3.13 μv, standard error = 2.2) relative to spiders and flowers in the neutral condition (mean = 9.31 μv, standard error = 2.9). The same main effect of emotion condition was found when peak amplitude instead of mean amplitude between ms was used as dependent measure, F(1,17) = 6.2, p = 0.023, partial η 2 =.27. In addition, there was a significant main effect of stimulus category on latency to peak of the Pb, F(1,17) = 5.05, p = 0.038, partial η 2 =.23. The Pb peaked earlier for flowers (mean = 275 ms, standard error = 9.6) than for spiders (mean = 295 ms, standard error = 8.3). No other significant main effects or interactions involving emotion condition or stimulus category were found for mean or peak amplitude or for latency to peak of the Pb, all psn 0.1. When considering only the spider condition, we also found a significant main effect of emotion condition, F(1,17) = 8.6, p = 0.009, partial η 2 =.34. Mean amplitude of the Pb was more positive for spiders in the neutral condition (mean = μv, standard error = 3.8) compared to the fearful condition (mean = 3.47 μv, standard error = 2.5). The same main effect of emotion condition was found when peak amplitude instead of mean amplitude between ms was used as dependent measure, F(1,17) = 9.48, p = 0.007, partial η 2 =.36. No other significant effects involving emotion condition were found, all psn 0.5. When considering only the flower condition, no significant effects involving emotion condition were observed, all psn 0.1. Mean amplitude of the Pb was similar in the neutral condition (mean = 5.84 μv, standard error = 3.7) and in the fearful condition (mean = 2.8 μv, standard error = 2.5) Discussion Results of Experiment 1 provide support for the notion that infants at nine months of age are inclined to learn socially about the valence of objects. Infants responded with increased attention as indicated by Nc amplitude to objects that were cued by a fearful as compared to a neutral face. This finding is in line with previous studies reporting increased Nc responses for objects associated with fear or disgust in infants at different ages (Carver & Vaccaro, 2007; Hoehl & Striano, 2010; Hoehl, Wiese, et al., 2008). There is some evidence that the effect of an adult's emotional expression on infants' attention to objects is to some extent dependent on the object category by nine months of age. Studies finding the same effect as the current experiment in younger infants (3- and 6-month-olds) used harmless toy objects (Hoehl & Striano, 2010; Hoehl, Wiese, et al., 2008), whereas a study on the effect of disgusted expressions in slightly older infants (12-month-olds) used potentially threatening objects (e.g., a remote-controlled dragon and spider, Carver & Vaccaro, 2007). In contrast, a previous study using toy objects with 9-month-olds reported no increased Nc amplitude for toys cued by a fearful face (Hoehl & Striano, 2010). In the present experiment we found a main effect of emotion condition, but the effect was greater for spiders than for flowers as evidenced by the hypothesis-driven post-hoc analyses. However, the interaction of emotion condition and stimulus category did not turn out to be significant. Thus, results of the separate analyses for both stimulus categories need to be interpreted with caution. It is possible that in this within-subject experiment the slight enhancement of Nc amplitude in the fearful-flower condition may represent a carry-over effect from the spider condition as infants may have been especially

6 S. Hoehl, S. Pauen / Evolution and Human Behavior 38 (2017) Fig. 3. ERPs time-locked to the onset of the flowers in Experiment 1. Infants showed no difference in brain responses for the flowers in the neutral vs. fearful condition. Negative is plotted upwards. alert due to the intermixed presentation of spiders and flowers and/ or may have mistaken some of the perceptually matched flowers for spiders. In addition to the effect on Nc amplitude we found a significant effect of emotion condition in the earlier Pb time-window. Reduced positive amplitude was observed for stimuli in the fearful face condition as compared to the neutral face condition. Again, this effect was mainly driven by the spider condition, but there was no significant interaction between emotion condition and stimulus category. The Pb is thought to reflect ease of stimulus processing and contextual updating with smaller positive amplitude indexing increased processing difficulty (Karrer, Karrer, Bloom, Chaney, & Davis, 1998). For instance, in 10-month-olds this component is larger for a familiar face than an unfamiliar face (Webb, Long, & Nelson, 2005), and it sometimes increases in amplitude across trials when stimuli are presented repeatedly (Nikkel & Karrer, 1994). Increased negativity in the Pb-Nc complex (i.e., decreased Pb amplitude and increased Nc amplitude) is typically found for fearful as compared to happy faces in infants with waveforms very similar to the ones observed in the present study (Grossmann et al., 2011; Nelson & de Haan, 1996). In accordance with this literature we interpret the increased negativity found in the fearful condition as indexing increased attention and increased processing demands for stimuli associated with a fearful face. Some limitations of Experiment 1 should be noted. As in previous studies on preparedness theory, we contrasted spiders with flowers (Rakison, 2009). However, as Quinlan (2013) pointed out with regard to visual search tasks, it is important to contrast responses to animals representing an ancestral threat with responses to other animals for which this is not the case in order to avoid a potential confound of superordinate category membership (animal vs. no animal). Furthermore, we did not assess infants' a priori familiarity with the stimulus categories. It is possible that infants are more familiar with flowers as compared to spiders and therefore more readily associate the less familiar target stimuli with fear. We conducted Experiment 2 in order to compare infants' responses to snakes, a second stimulus category representing an ancestral danger with perceptually matched fish, i.e., another animal category. In addition, we asked parents to rate infants' prior experiences with both categories. Experiment 2 was identical to Experiment 1, except that perceptually matched snakes and fish were presented instead of spiders and flowers (see Fig. 1 for stimulus examples). 3. Experiment Method Participants In Experiment 2, 20 infants were included in the sample (7 females, age range: 278 days to 300 days; average age: 290 days). Another 26 infants were tested but excluded from the sample because they failed to reach the minimum requirement of 10 artifact-free trials per condition for averaging. Infants that were excluded from the sample had seen a similar amount of trials in each condition, i.e., they did not drop out specifically because they avoided looking at snakes. Each infant contributed 10 to 26 trials in the fearful snake condition and 10 to 28 trials in the neutral snake condition. Furthermore, each infant contributed 11 to 24 trials in the fearful fish condition and 10 to 27 trials in the neutral fish condition. An average of 16 trials was contributed per infant per condition in each of the four conditions. Before testing, infants' parents were asked whether and in what context their infants had ever seen or regularly see snakes and fish.

7 410 S. Hoehl, S. Pauen / Evolution and Human Behavior 38 (2017) Results Parental report of infants' prior experiences Of the 20 infants that were included in the final sample 9 were rated to have seen fish regularly. The majority of these infants (N = 6) had toys looking like fish, 2 had seen pictures of fish (books, stickers), and 1 family had an aquarium at home. An additional 3 infants were rated to possibly have seen fish occasionally. In contrast, only 3 of these infants had snake-shaped toys and 1 additional infant was rated to have seen snakes at one occasion (i.e., during a visit to the zoo) ERP results ERPs for Experiment 2 are shown in Figs. 4 and 5. An enhanced negativity is visible for snakes as compared to fish in both emotion conditions. The same statistical analyses for mean amplitude of the Nc between ms were carried out as in Experiment 1. Withinsubject factors were emotion condition (fearful, neutral), stimulus category (snake, fish), and electrode site (F3, C3, FC1, FC2, FZ, CZ, F4, C4). In further analyses peak amplitude and latency to peak of the Nc were analyzed. As in Experiment 1, mean and peak amplitude and latency of the Pb between ms were also analyzed, but no main effects or interactions involving the factors emotion condition or stimulus category were found, so no results regarding the Pb will be reported Nc: ms A significant main effect of stimulus category was found for mean amplitude of the Nc, F(1,19) = 8.38, p = 0.009, partial η 2 =.31. Amplitude was greater for snakes (mean = 9.45 μv, standard error = 2.2) than for fish stimuli (mean = 4.76 μv, standard error = 2.1), suggesting that 9-month-olds direct more attention towards snakes versus fish irrespective of the emotional context. There was no significant main effect of emotion condition, p = 0.647, and no interaction of stimulus category by emotion condition, p = The same main effect of stimulus category was found when peak amplitude instead of mean amplitude between ms was used as dependent measure, F(1,19) = 14.29, p = 0.001, partial η 2 =.43. No other significant main effects or interactions involving emotion condition or stimulus category were found for mean amplitude or peak amplitude and no effects were found for latency to peak of the Nc, all psn Nc effects in infants who have never or only rarely seen fish before We conducted additional analyses with a subgroup of 11 infants who were reported to have never (N = 8) or only occasionally (N = 3) seen fish before. A significant main effect of stimulus category was found for mean amplitude of the Nc, F(1,10) = 5.75, p =0.037,partial η 2 =.36. Amplitude was greater for snakes (mean = μv, standard error = 2.8) than for fish stimuli (mean = 7.36 μv, standard error = 3.0). The same main effect of stimulus category was found when peak amplitude instead of mean amplitude between ms was used as dependent measure, F(1,10) = 8.12, p = 0.017, partial η 2 =.45. No other significant main effects or interactions involving emotion condition or stimulus category were found for mean amplitude or peak amplitude and no effects were found for latency to peak of the Nc, all psn Discussion As opposed to spiders, which elicited increased attention only after being paired with a fearful vs. neutral face, snakes received more attention than fish regardless of the emotional face context. This finding was not predicted by the preparedness account as snakes should not elicit fear or heightened attention prior to any direct or vicarious learning experience. It is also not compatible with the social learning assumption Fig. 4. ERPs time-locked to the onset of the fearfully cued targets in Experiment 2. Infants showed increased Nc amplitude in response to snakes as compared to fish targets regardless of the emotional expression of the face. Negative is plotted upwards.

8 S. Hoehl, S. Pauen / Evolution and Human Behavior 38 (2017) Fig. 5. ERPs time-locked to the onset of the neutrally cued targets in Experiment 2. Infants showed increased Nc amplitude in response to snakes as compared to fish targets regardless of the emotional expression of the face. Negative is plotted upwards. that would predict increased attention for any stimulus associated with fear. Rather, increased attention for snakes is in line with nonassociative accounts of fear acquisition. In addition to ERPs, we assessed infants' prior experience with the stimulus categories based on parental reports. The majority of infants taking part in this study were rated to have seen fish previously and almost half of them were rated to have seen fish regularly, e.g., as toys or in picture books. In contrast, only a fifth of all infants were rated to have seen snakes occasionally or regularly. Although it is unlikely that infants had a priori experiences with the particular exotic fish we used as stimuli here, most of them had seen items of the same category of animals before. Thus, the possibility that infants simply responded with increased attention to the less familiar category had to be considered. In an additional analysis we therefore included only infants with no or very limited prior experience with fish. We observed the same main effect of stimulus category as in the whole sample. Consequently, more prior experience with fish compared with snakes cannot fully account for the effect of stimulus category on Nc amplitude in the current study. In contrast to our current findings, Deloache and LoBue (2009) found no differences in looking times when 9- to 10-month-olds watched films of snakes as compared to other exotic animals (they did not report infants' prior experiences with the tested animal categories). The advantage of using ERPs, and in particular the Nc, is that differences in the amount of attention directed to visual stimuli can be assessed when infants look at these stimuli while showing no overt differences in their behavioral reactions (Reynolds & Richards, 2005). Therefore, at least in some cases, ERPs have proven to be a more sensitive measure than looking times (de Haan & Nelson, 1997; Hoehl, Reid, et al., 2008; Snyder, 2007). Nevertheless, it will be important to replicate the current findings to rule out that our specific set of snake and fish stimuli (although perceptually matched) elicited differences in attention. 4. General discussion In the current study we found partial support for social learning and preparedness theory as well as for non-associative fear acquisition in early human development. Nine-month-old infants directed increased attention towards spiders and flowers after these were gaze-cued by an adult with a fearful facial expression, though the effect was more clear-cut for spiders as compared to flowers, as predicted by the preparedness account. Snakes, in contrast, elicited greater Nc amplitude than fish regardless of the emotional expression of the face looking at the target stimulus. The latter finding is in line with a non-associative account of fear acquisition. Taken together, results speak for content specificity of the fear system in early human ontogeny. According to Rachman's theory, specific phobias can be acquired through direct conditioning, observational learning, and verbal instruction (Rachman, 1977). As an additional pathway of fear acquisition, Poulton and Menzies (2002) suggested a non-associative fear acquisition account. In this view, certain fears have evolved in phylogeny and do not require any learning experiences during ontogeny. Accordingly, specific phobias result from a failure to habituate to these threats during development. How does the existing ERP evidence on infants' fear associations fit with these theoretical frameworks? As shown in previous work, very young infants quickly associate fear-expressions with toys which are clearly not threatening or resembling ancestral threats for humans (Hoehl, Wiese, et al., 2008). This effect has been found in 3-month-olds (Hoehl, Wiese, et al., 2008) and 6-month-olds (Hoehl & Striano, 2010). Interestingly, however, 9-month-old infants did not show the same effect when harmless toys were coupled with fearful faces (Hoehl & Striano, 2010). It seems that fear associations are easily built even for nonthreatening objects very early in development, but not somewhat later within the first year after birth. By the age of 9 months, infants

9 412 S. Hoehl, S. Pauen / Evolution and Human Behavior 38 (2017) have acquired several months of experiences with the category of toys and do no longer respond with increased attention to toys after seeing a fearful face looking at these highly familiar stimuli (Hoehl & Striano, 2010). This, of course, does not imply that infants have a disposition to fear toys that they habituate to until 9 months of age. Rather, these results suggest that humans may be more prone to making fear associations with arbitrary objects through observational learning in the first months, whereas highly familiar, non-threatening objects cannot be as easily associated with a fearful expression later on. Regarding objects that represent an ancestral danger, the nonassociative account of fear acquisition suggests that they should elicit a fear response even in very young infants. It is currently unclear whether this is the case, but the fact that infants quickly detect snakes and spiders in visual displays (LoBue & DeLoache, 2010; Rakison & Derringer, 2008) may support this view. Furthermore, in the current study 9-month-olds directed enhanced attention to snakes as compared to fish regardless of the emotional expression of the accompanying face. This finding may support the notion that snakes induce fear responses in infants, or at least heightened alertness, and that the infants in our sample have not yet habituated to snakes. This is in line with parental reports of only very limited exposure to snakes or snake-like stimuli in this age group. Evolved fear responses may also exist for spiders, but this is not reflected in generally increased attention for spiders as compared to flowers in 9-month-old infants. One possibility is that infants in our sample have had more early exposure to spiders than snakes and consequently have already habituated to the sight of a spider. However, they may still be more prone to building fear associations with spiders than with non-threatening objects, i.e. to re-learn this fear. In this view, evolutionary acquired fears that do not require learning are normally habituated to after early exposure without experiencing harm from these ancestral threats. Nevertheless, these fears may be relatively easy to rekindle later on, for instance, through observational learning. Thus, the different pathways of fear acquisition may not be independent from each other, but rather highly intertwined. Preliminary support for this notion comes from Experiment 2. Only 4 of the 20 infants participating in Experiment 2 were reported to have seen snakes or snake-like objects before. This subsample was too small to conduct statistical analyses. Notably, though, Nc amplitude in this subsample was particularly large in response to snakes associated with fear (mean = 19,72 μv) as compared to the other conditions (means between 2,95 and 5,17). This can be seen as preliminary evidence that infants who have already habituated to snakes through early exposure do not show heightened attention to this category per se, but only when snakes are associated with fear. Experiences with snakes should be manipulated in further research in order to test this assumption more directly. To date, there is conflicting evidence regarding the selectivity of observational fear learning for objects that are potentially threatening in infants around one year of age. In some studies effects of fear expressions on infants' avoidance of unfamiliar objects were found for unthreatening toys (Mumme & Fernald, 2003), whereas in other studies avoided target objects were described as threatening (Carver & Vaccaro, 2007; Klinnert, Emde, Butterfield, & Campos, 1986). Sometimes larger effects were observed for ambiguous or threatening objects than for un-threatening objects (Gunnar & Stone, 1984). In general, it seems that by the end of the first year infants avoid novel objects associated with the intense negative emotional reaction of an adult even if these objects are not inherently threatening. However, infants may somewhat more readily associate stimuli with fear that represent an ancestral danger of humans, as shown for spiders as compared to flowers in the present study in 9-month-olds. The same seems to be true somewhat later in development for snakes (Deloache & LoBue, 2009; Rakison, 2009). This may be after infants were exposed to snakes (e.g., snakelike toys or pictures of snakes in books etc.) and have habituated to them, so that they do no longer respond to the sight of a snake per se. There are some limitations to the current study. For instance, it might be interesting to test the effects of gaze direction in the current paradigm. Eyes of the models were always directed at the spiders and flowers. Based on previous findings (Hoehl & Striano, 2010; Hoehl, Wiese, et al., 2008), we predict that no effects of emotion condition will be found if eyes are not directed towards the targets. This would support the assumption that infants make use of social cues provided by adults when directing their attention towards potential threats in the environment. The main effect of stimulus category in Experiment 2, however, should be independent of the adult's eye gaze. Furthermore, it would be interesting to conduct prospective research with infants at risk for autism, a disorder characterized by deficits in the use of social cues such as gaze (Hoehl et al., 2009). Future studies should also investigate individual differences in infants' attention direction in response to fearful expressions. In adults, fearful facial expressions lead to increased gaze cueing effects even for neutral targets in highly anxious subjects (Mathews, Fox, Yiend, & Calder, 2003). In infants, individual differences in attention disengagement from fearful faces have been linked to genetic variations affecting the serotonin system (Leppänen et al., 2011). Investigating individual differences in emotion processing in infancy may potentially provide valuable information on the development of anxiousness and anxiety disorders. In fact, individual differences, both in terms of prior experiences and temperament, may have contributed to the different results obtained in our two samples in the current study. We did not assess infants' prior experiences with the target categories in Experiment 1, so the notion that infants may have already been exposed to spiders but not snakes is rather tentative. In addition, further studies should be conducted using a larger array of stimulus categories, including landdwelling non-threatening animals. To conclude, we present evidence that 9-month-old infants direct increased attention towards spiders that were gaze-cued by an adult with a fearful vs. neutral facial expression. In line with preparedness theory, this effect was less pronounced for flowers that were perceptually matched to the spiders. Snakes received a higher amount of attention than perceptually matched fish regardless of the emotional expression of the face, thus supporting non-associative accounts of fear acquisition. In sum, this pattern of results suggests that different pathways of content dependent fear-acquisition exist in infancy that might also be interrelated: Previously habituated fears of ancestral threats may later be especially easy to rekindle through conditioning or observational learning. This may lead to quicker and more robust fear associations for these stimuli, at least in some individuals. Further research is needed that focuses on inter-individual differences in infants' responses to different stimulus categories and infants' sensitivity to social emotional cues. The combination of these two factors may play a key role for explaining the development of object-specific fears. Acknowledgements We are grateful to the infants and parents who participated in the experiments and to the student research assistants who carried out the data collection. This work was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG) [grant number HO 4342/2-2]. References Carver, L. J., & Vaccaro, B. G. (2007). 12-month-old infants allocate increased neural resources to stimuli associated with negative adult emotion. Developmental Psychology, 43(1), ( pii). Cook, M., & Mineka, S. (1989). Observational conditioning of fear to fear-relevant versus fear-irrelevant stimuli in rhesus monkeys. Journal of Abnormal Psychology, 98(4), Cook, E. W., Hodes, R. L., & Lang, P. J. (1986). Preparedness and phobia: effects of stimulus content on human visceral conditioning. Journal of Abnormal Psychology, 95(3), de Haan, M. (Ed.). (2007). Infant EEG and event-related potentials. Hove, England: Psychology Press.