When anthropomorphism is not as much of an automatic process as we think: a revised Heider and Simmel fNIRS stu

(1)

ANTHROPOMORPHISM AND THE PROCESS OF IMAGINATION

When anthropomorphism is not as much of an automatic process as we think: a revised Heider and Simmel fNIRS study.

Supervisor: dr. A. B. Satpute, Northeastern University

Co-assessor: dr. S. Ghebreab Word count:

No. of References:38

Master Brain and Cognitive Science, University of Amsterdam

Track: Cognitive Sciences University of Amsterdam Gabrielle.davelaar@gmail.com

(2)

Abstract

Anthropomorphism is defined as the act of to attributing human actions and emotions to objects that are unable of that type of action or emotion. It is generally believed that

anthropomorphism is an hardwired automatic ability which arises from the need of socialization. However, some cases such as with people with autism do not anthropomorphize or too a lesser extent. This research shows that anthropomorphism doesn’t occur automatic but rather comes from a learning curve often created from cueing or priming. Participants started to use more words associated with anthropomorphism when a questionnaire beforehand used prime words. Moreover, no correlation was found between emotional intelligence and the amount of words associated with anthropomorphism contradicting earlier findings.

(3)

INTRODUCTION

“I’m sorry, Wilson. Wilson, I’m sorry! I’m sorry! - Cast Away

In 2000, the movie Cast Away showed the life of a stranded man on an island and the volleyball he found in the flight wreck. The volleyball was called Wilson and the stranded man, Chuck Noland, considered him as a companion and friend in need. The inanimate object was the one thing that made Chuck not go crazy on the island. Perceiving people as having thoughts, feelings, and motivations and to be able to predict these mental states, also known as theory of mind (ToM), is important (Premack & Woodruff, 1978). This ability is critical for building social relationships, which promotes a happy life and subjective well being (Aristotle, 2300 BC; Diener & Seligman, 2002). Tomasello et al. (1993; 2008) argued that ToM is what makes humans unique from other creatures on this world. Apes are capable of mentalizing what others feel or perceive but in lesser degree than humans. Indeed, it has been proposed that culture and language both depend on ToM to begin with (Tomasello et al., 2008).

Humans are not only capable of perceiving feelings, motivation and other mental states only in other humans; we tend to also see them in a variety other living and nonliving objects though these objects are unable to take such action. This tendency is referred to as anthropomorphizing, in which humans project thoughts, feelings, and beliefs to animals (Serpell, 2003, Watts, 2006), imaginary entities (Epley et al., 2008) and even abstract objects ( Heider and Simmel, 1944). According to Guthrie (1995), the reason why anthropomorphism occurs is that it is an innate system which guides behavior in order to survive in a social and complex environment. Research has shown that without social interaction, humans grow unhappy and the risk factors from isolation can increase many health issues and even causes the brain to rewire itself (Cacioppo et al., 2009; Cacioppo & Hawley, 2009; Lieberman, 2013) As suggested by the movie Cast Away, in the absence of actual people, Chuck Noland survived because he anthropomorphized “Wilson”, the volleyball.

Indeed, it is not uncommon for people to see human like qualities in abstract objects. In the classic 1944 study of Heider and Simmel, participants were able to automatically imbue

(4)

social meaning in three moving geometric shapes; a large triangle, a small triangle and a small circle. Not only did participants see intentions in the two minute clip of the moving shapes, but also gender, relationships, personal traits and emotions (Heider & Simmel, 1944). It would appear that anthropomorphism occurs with 4-5 year olds’ and even more so than older people (Carey, 1985; Berry & Springer, 1993). Several researchers have indicated that anthropomorphism is automatic ( Heider & Simmel, 1944; Wheeler & Petty, 2001; Chartrand, 2008). It seems that personality traits (Chartrand, 2008), faces (Duffy, 2003; Aggarwal & McGill, 2007) and human like movement (Visch & Tan, 2009; Morewedge, et al., 2007). The notion that people automatically, reflexively, or effortlessly perceive even abstract shapes as having thoughts and feelings has lead to several lines of work that focus on to identifying the physical characteristics of the stimuli that trigger anthropomorphism. For example, it has been suggested that anthropomorphism occurs when motion and speed are on the same level as humans move ( Morewedge et al., 2007) or when facial features are added to the object (Duffy, 2003; Aggarwal & McGill, 2007).

However, a few lines of work suggest that it may not be the case. Work of Heberlein et al. (2006) found that anthropomorphism also depends on the way the questions are asked, implying that conceptual structures also play a role. The same observation was done by Ganea et al. (2014) who ran a series of experiments on young children. The children were given information about animals in two ways; an anthropomorphic description and a factual description. The children were more likely the characterize human attributes to the animals from stories where anthropomorphic descriptions were used. Moreover, research coming from consumer research also showed that there is a link between anthropomorphism and priming (Firtzsimons et al., 2008; Aggarwal & McGill, 2012), suggesting again that anthropomorphism is flexible rather than automatic, reflexive, and inflexible. Lastly the age of acquiring anthropomorphism seems to differ, where urban children develop it around 3 and 5 years old and older in other groups (Herrmann et al., 2010) and correlates with the exposure to media (Waxman et al., 2014).

Given this discrepancy in the literature, we more closely examined the initial Heider Simmel investigation which lead to the notion that anthropomorphism is an automatic human

(5)

tendency. A closer inspection of that seminal work provides a few reasons to suspect that the methodology and tasks demands in the study may have oriented participants to think of the stimuli in an anthropomorphic way to begin with, rather than testing whether anthropomorphism occurred actually occurred spontaneously and ubiquitously across (neuro)typical participants. For example, participants were instructed to answer questions such as what kind of a person is the big triangle? Or, why did the circle go into the house? They were also limited to female psychology participants. Here, we examined whether there was more variability in anthropomorphic tendencies when participants were provided with these “nudges” so to speak to interpret the stimuli in this way to begin with. Quite simply, participants viewed the initial Heider Simmel video and provided verbal descriptions of what they saw. We then measured the degree to which their descriptions involved anthropomorphization of the abstract shapes presented in the video by counting how many social words were used in their descriptions. We hypothesized that while some participants might spontaneously anthropomorphize the video, others might not, and that this would be reflected in their verbal descriptions. Critically, all participants then answered the sequence of guided questions from the initial Heider Simmel study, and the provided a description of the video again. We then examined whether answering these questions increased the degree to which participants provided anthropomorphic descriptions.

It is, of course, possible that participants may automatically perceive the stimuli in anthropomorphic ways, but nevertheless describe them in non-social ways. This would suggest that people do autonomically anthropomorphize stimuli even if their verbal descriptions do not suggest they do the first time around. To get a converging evidence as to the extent to which people spontaneously anthropomorphized while watching the video, we turned to findings in neuroscience. These findings suggest that the anterior medial prefrontal cortex is frequently engaged when anthropomorphizing. Though results from other studies focusing on anthropomorphism vary. In general the areas identified are known to be activated during metalizing such as an area in the anterior temporal lobes close to the amygdala and the temporoparietal junction (TPJ) and the angular gyrus (Castelli et al., 2000, 2002; Waytz et al., 2010). Moreover, Cullen et al. (2014) concluded that the grey matter volume in the TPJ is a

(6)

predictor whether people are more easily engaged in the process of anthropomorphism while using fMRI. In this article, we focus on a framework where we use fNIRS to capture the brain activity in the prefrontal cortex (PFC), while being engaged in an anthropomorphic task. Reason for this is the large body of work concerning the strong correlation between activity in the PFC and the affective-motivational expression ( Aranyi et al., 2015). Moreover, fNIRS offers a better temporal resolution compared to fMRI and a better robustness and signal stability to motion artifacts than in EEG and is very suitable when investigating the PFC (Ayaz et al., 2011). With this experimental setting we ought to find out whether anthropomorphizing is an automatic process not fueled by word priming. Our objective here to use fNIRS is to improve the insights into the process of language, anthropomorphic thought and affective expression by monitoring the different phases of the experiment where all three play a significant role.

METHOD SECTION

Subjects

Twenty nine healthy consenting undergraduate students participated in the study for course credit (6 males, 23 females, mean age= 18.63, age ranging from 17 to 20). Participants were coming from different races ( White or Caucasian, black or African American, Asian, Hispanic-latino, ) and social backgrounds.

Affective stimuli

The original Heider and Simmel animation was shown, which shows three geometrical shapes (a large triangle, a small triangle and a circle) moving in and around a rectangle in ways that can be described with social behavioral words (e.g. ‘fighting’,’playing’, etc.; see: http://www.youtube.com/watch?v=wp8ebj_yRI4).

fNIRs set up

Prefrontal cortex activation was measured with the NIRSport (NIRx Optical Neuroimaging, Medizintechnik GmBH, Berlin, Germany). The continuous-wave multi-channel

(7)

system uses 2 wavelengths at 760 and 850 nm to measure concentrations of oxygenated (oxy-Hb) and deoxygenated (deoxy-Hb) hemoglobin. The NIRScap consisted of 8 optodes ( 4-source/4-detector) in a 4-channel arrangement with a 3 cm spacing between the optodes. The following EEG electrode system areas were used: Fp1, F3, F7, Fp2, F4, F8. Optical signals were between 2.5Hz and 62.Hz and data from this was later used to calculate oxy-Hb and deoxy-Hb changes based on a modified Beer-Lambert law.

Experimental design

The experimental design involved modifications to the original Heider Simmel Study to allow for examining how anthropomorphic word use shifts before and after being provided with social primes in questions.

The research consisted of four phases. In phase 1, participants were shown the video and were instructed to passively watch the screen. It was explicitly mentioned to watch it passively due to confusion by participants in the first two test rounds at the beginning of the experiment. Following by phase 2 in which participants were asked to describe the video while they watched it. In phase 3 participants were asked to describe the video from memory. Lastly, describing the video (duration 1:27 minutes) from memory after being cued with the same questions from the original Heider Simmel study. These included the following questions:

1. What kind of a person is the big triangle? 2. What kind of a person is the little triangle? 3. What kind of a person is the circle?

4. Why did the two triangles fight? 5. Why did the circle go into the house?

6. In one part of the movie the big triangle and the circle were in the house together. What did the big triangle do then? Why?

7. In one part of the movie the big triangle was shut up in the house and tried to get out. What did the little triangle and the circle do then? Why did the big triangle break the house?

(8)

participant, obtained a brief (5 seconds) calibration for estimating gain levels and a two minute recording of fNIRs recording while at rest during which subjects were instructed to relax and to look at the monitor.

Upon completion, participants were debriefed about the study containing information about the study and contact information in case more information is wanted. Study procedures were approved by the Northeastern University Institutional Review Board.

Data analysis

Audio scripts were transcripted by two people. Errors were verified by a third person. We specifically looked at ‘social/non-social’ words that were used in the original questionnaire from Heider and Simmel. In the figure below an overview of the social/ non social words are displayed. The data collected was both non automated/automated counted, meaning for the non automated two people counted both categories for all participants and for the automated part LIWC2007 was used. No significant differences were found except for some cases which were small spelling errors. Moreover, the total amount per phase of word usage was also calculated (both non automated and automated). These were used to measure a percentage of the word use in every phase per category.

As for the the emotional intelligence dataset, the counting of the words was done via a Python 2.5 script using the Pandas library and Plotly. A random manual check was done to check for errors. None were found.

Pre-processing

The pre-processing consists of several stages. First by removing physiological noise and slow drifts using low-pass and high pass filtering from the SPM nirs matlab package (http://www.fil.ion.ucl.ac.uk/spm/). Optical signal is converted using the modified Beer-Lambert Law into HbO and HbR signals. Moreover, only channels that were approximately 2.5-3 cm apart from each other were included in the analysis.

(9)

Result 1 _{: The first time participants started using social words started out small 10 out of} 29 participants started using social words in phase one, whereas the final phase showed the highest rate of first usage of social words. (16 out of 29). This would suggest that the questions are cueing participants to use more social words, but it isn’t necessarily influencing the anthropomorphic process. We reasoned that if so, many participants would still use social words in phase 1, just fewer in the last phase. Only one participant didn’t use any social words during the whole experiment.

Result 2: A repeated measure anova was conducted and the results show that there was a significant effect how many social words were used by participants in a specific phase F(1.562, 42.185) = 16.301, p <.0005. There were some significant correlations, which were all negative correlated indicating that the object words are influenced by how many social words are used. Five significant results were found between object words versus social words. Object words versus social words in phase 1 (r = -.486, p (two-tailed) < .01.), indicating that if participants used social words in the first phase they would use less object words. This effect can also be found in phase two (r = -.569, p (two-tailed) < .01) and in phase three (r = -.455, p (two-tailed) < .05.) Moreover a significant result was found between object words in phase 2 versus social words in phase 1 (r = -.455, p (two-tailed) < .05.) and between object words in phase 2 versus social words in phase 3 (r = -.397, p (two-tailed) < .05.). This indicates that even previous phases have influence on the next phase on how much objects words are used versus social words.

Result 3 _{: Individual differences. Next we examined whether some people tended to use} social words more in phase 1 and 2 also did so in phase 3. To examine this, we conducted a factor analysis. If people are one-dimensional, you should only see one factor. We found two. One corresponded to people’s spontaneous use of object/person words prior to being cued. The second corresponded to people’s cued use of object/person words.

The method used for this was a PCA with varimax rotation (eigenvalues >1) accounting for a total of 69% of the various. Moreover, the sampling was sufficient based on KMO and Bartlett test (KMO =.619, Bartlett’s test of sphericity χ2 (15) = 54.975, p <.001.).

(10)

and social word use and were unable to find any correlations (Pearson’s r (201) = .021, _{p .87) nor} were found to sexe (figure 1a (female): Pearson’s r (75) = .019, _{p .9) ( & 1b (male): Pearson’s r} (125) = .074, _{p .6)).}

Figure 1. Correlation between emotional intelligence and social word usage (a. female, b. male).

To summarize the behavioral results shortly, there is an effect of cueing and non social words are influencing the amount of social words in both the same and the phase after. One participant didn’t use any social words which could be due to many reasons. Moreover, our factor analysis showed two dimensions; one which corresponds to people’s spontaneous use of object/ social words and a second one corresponding to people’s cued use of object/social words. We also confirmed that there is no correlation between emotional intelligence and social word usage.

Neuroimaging data

The baseline had a total of 29 participants, the amount of subjects contributing to 5x6 channel grid. The number of participants contributing to the first stage was 29 with a minimum of 12 and a maximum of 28 per optode. A general linear model was performed together with a boxcar regressor for the duration of the video period convolved with a canonical hemodynamic response function from the SPM software package. For 1 participant the GLM failed to converge, this participant was excluded from further analyses.

(11)

(02HB) nor the deoxyhemoglobin (HHB). This indicates that both the tissue’s inflow of oxygen and the absorbed amount of oxygen by the tissue are not different from the baseline concentrations. Figure 2 shows the correlations from phase 1 and phase 2. No significant P_{-values were present. All graphs contain 30 sensors, though only 27 sensors which contain} actual histograms. The other three were not capturing significant signals (ranging from SNR due to spacing of optodes to detectors or the lack of skin contact with the optodes and detectors due to hair or poor fit of the cap).

(12)

(13)

Figure 2 (a, b, c). The forehead layout of the figures are for all three the same where dorsal is at the top,

ventral at the bottom, left on the left side, right on the right side etc. Figure 2.a values of oxy in both baseline (blue) and phase 1 (red).Graphs presenting only equal red and blue color lacked significant input data.No significant _P-values were present. Figure 2.b values of deoxy in both baseline (blue) and phase 1 (red). Graphs presenting only equal red and blue color lacked significant input data. No significant

P-values were present.

Figure 2.c values of total HBOT in both baseline (blue) and phase 1 (red). Graphs presenting only equal red and blue color lacked significant input data. No significant _P-values were present.

CONCLUSION

The question motivating this study is whether anthropomorphism is an automatic process. We examined people’s narratives while they watched a video involving abstract shapes, and also used ambulatory imaging to record activity in prefrontal cortex while they spoke.

According to our data anthropomorphism doesn’t occur automatic, participants tend to use object words more frequently in the first two phases than in the third phase. We hypothesized that the questionnaire primes responses in the third phase, which seems the case since one of the primary words in the social processes category, namely house, is shown mainly in the third phase. Where participants first used mostly rectangle to name the figure in the first and second phase, house is used most in the third phase instead of rectangle. It could well be the case that participants use anthropomorphic words due to the questions, after all house is used in the questions and participants are almost forced to see a human being in the figures due to the questions. Interestingly, the descriptions the participants give while answering questions about what kind of person the figures are, are not found back in the final description of what happened in the movie clip. This line of reasoning, can also found back in the research of criminology, where wrongful convictions mainly happens due to wrong testimonies either by wrong recall or due to psychological “set”. The concept “set” is referred to the process prior a line up where an officer could make a comment that could interfere with the witness recollection of the event, such as “we caught a guy who did a similar crime, is this the same person who attacked you?” (Douglass, 2006).

(14)

would assume that this effect is shown in the amount of participants using anthropomorphistic words in phase one. However, the majority of the participants in this study showed that when it comes to object words versus social processes that this is not the case. Besides that, there are known cases of people in which anthropomorphism does not occur, such as people with bilateral amygdala damage (Heberlein, 2004) or autistic spectrum disorder (Castelli et al., 2000). It is often speculated that the reason behind this, is the fact that people with autistic spectrum disorder lack the theory of mind (ToM), which is related to mentalizing the feelings of others (Frith and Frith, 2003).

Another note that should made is the fact that not every single object humans’ encounter are being humanized. There seems to be a selectivity (Waytz, 2010), how this works is not fully understood yet, but understanding how anthropomorphism works and more important when it doesn’t work, gives a valuable insight on how social perception works. It gives insights on how to treat socially disconnected people, how people make moral decisions and how people are capable of inhuman actions (Epley et al., 2008; Zimbardo, 2008). After all, the antonym for anthropomorphism is dehumanization, in which people can treat other people as non human entities.

We know from Neuroscience that the mPFC plays an essential area in social cognition ranging from story comprehension ( Mar, 2011), social reward (Fiske, 2006) mentalizing (Frith and frith, 2003) and various other social cognition abilities.

More interesting is the results from neuroimaging studies focused on dehumanization and objectification, which showed that the mPFC was not activated while people see other people as automatons (Fiske, 2006).

The non significant fNIRS results could have several reasons, foremost this study only had one trial with a small sample size(29) where the minimum of 12 participants per channels was reached and a maximum of 28, with an average of 23. Implicating that the results may be different in a larger trial due to the fact that almost one third of the participants had some issues while measuring the results. Adaptation of the measurement protocol could be beneficial for better results and thus more significant results. Factors such as long hair, curly hair and skin tone are known to play a role in achieving adequately signal to noise ratio. Previous studies have

(15)

shown that the amount of melanin in the skin affect the signal to noise ratio where participants with higher levels of melanin need higher source power levels in order to achieve adequate signal to noise ratios ( Lloyd-Fox et al., 2014). As for hair, dark-colored hair or a large amount of hair affect the amount and quality of near-infrared light passing through (Nishiyori, 2016). Making fNIRS research more suitable for infants or people with less hair.

To conclude, Heider & Simmel (1944) showed an interesting effect which proves that people are able to anthropomorphize abstract figures quite easily, however there is more to this effect than solely being automatic. Because what could appear as automatic may not be as automatic as we think. Humans are prone to anthropomorphize, as mentioned earlier in the introduction as a part of an innate system. However, it doesn’t happen in all cases naturally. There are outside influences that make people to anthropomorphize more in certain situations than in others.

Bibliography

Aristotle, Nicomachean Ethics (2004), ed. Hugh Treddenick. London: Penguin. Aggarwal, P., McGill, A. (2007) Is That Car Smiling at Me? Schema Congruity as a Basis for Evaluating Anthropomorphized Products. Journal of Consumer Research, 34(4).

Aranyi, G., Charles, F., Cavazza, M. (2015). Anger-based BCI using fNIRS

Neurofeedback. Proceedings of the 28th Annual ACM Symposium on User Interface Software and Technology. ACM, New York, NY, 511-521.

Asquith, P.J. (1986). Anthropomorphism and the Japaneses and Western Tradition in Primatology. Cambridge: Cambridge University Press.

Ayaz, H., Shewokis, P.A., Bunce, S., Onaral, B. (2011). An optical brain computer interface for environmental control. Conf. Proc. IEEE Eng. Med.Biol.Soc. 6327-6330.

Berry, D.S., Springer K.(1993) Structure, motion, and preschoolers' perceptions of social causality. Ecological Psychology, 5, 273-283

Cacioppo et al. (2009) In the Eye of the Beholder: Individual Differences in Perceived Social Isolation. Predict Regional Brain Activation to Social Stimuli. Journal of Cognitive

(16)

Neuroscience, 21 (1): 83 DOI: 10.1162/jocn.2009.21007

Cacioppo, J.T., Hawkley, L.C. (2009). "Perceived social isolation and cognition". Trends in Cognitive Sciences 13 (10): 447–54.

Carey S. (1985) Conceptual Change in Childhood. Cambridge: MA: MIT Press.

Castelli F, Happe F, Frith U, Frith C.(2000) Movement and mind: a functional imaging study of perception and interpretation of complex intentional movement patterns.

Neuroimage;12(3):314-25.

Castelli, F., Frith, C., Happe, F., Frith C., Autism and Brain Mechanisms for the Attribution of Mental States to Animated Shapes. Brain, 125, 1839-49.

Chartrand, T.L., Fitzsimons, G.M., Fitzsimons, G.J. (2008) Automatic Effects of Anthropomorphized objects on Behavior. Social Cognition, 26, 2, 198-207.

Cullen, H., Kanai, R., Rees, G. (2014). Individual Differences in Anthropomorphic attributions and Human Brain Structure. Soc Cogn Affect Neurosci, 9 (9): 1276-1280.

Diener, E., & Seligman, M. E. P. (2002). Very happy people. Psychological Science, 13,80–83.

Douglass, A.B., Steblay, N. (2006). Memory distortion in eyewitnesses: a meta-analysis of the post-identification feedback effect. Appl. Cognit. Psychol., 20: 859–869. doi:

10.1002/acp.1237

Duffy, B.R. (2003). Anthropomorphism and the social robot. Robotics and Autonomous Systems. 43, 177-190.

Epley, N., Waytz, A., Akalis, S., Cacioppo (2008) When we Need a Human: Motivational Determinants of Anthropomorphism. Social Cognition. 26(2): 143-155.

Epley, N., Akalis, S., Waytz, A., Cacioppo, J.T. (2008) Creating Social Connection Through Inferential Reproduction. Loneliness and Perceived Agency in Gadgets, Gods, and Greyhounds. Psychological Science. 19 (2): 114-121.

Ganea, P.A., Canfield, C.F., Simons-Ghafari, K (2014) Do Cavies talk? The Effect of Anthropomorphic Picture Books on Children’s knowledge about animals. Front Psychol. 5(283).

(17)

Gilroy S.W., Porteous, J., Charles F., Cavazza, M., Soreq, E., Raz G., et al. (2013) A Brain-computer interface to a plan-based narrative. Proceeding of the 23rd International Joint Conference on Artificial Intelligence (Beijing).

Herrmann, P., Waxman, S.R., Medin, D.L.(2010) Anthropocentrism is not the first step in children’s reasoning about the natural world. Proc. Natl. Acad. Sci. U.S.A. 107

Frith U., Frith C.D.(2003). Development and neurophysiology of mentalizing. Philosophy Transcripts Royal Society London Biology of Science:459-73.

Guthrie,S. E. (1995) Faces in theClouds:A New Theory of Religion(Oxford Univ. Press,Oxford)

Heberlein, S., Adolphs, R., McGaugh, J.L.(2004). Impaired Spontaneous

Anthropomorphizing despite Intact Perception and Social Knowledge. Pnas 101(19), 7487-7481. Heider, F., Simmel, M. (1944). An experimental study of apparent behavior. American Journal of Psychology, 57, 243–59

Lieberman, D. (2013) Social:Why our brains are wired to connect (OUP Oxford, Oxford) Premack, D., & Woodruff, G. (1978) Does the chimpanzee have a ‘theory of mind’? Behavioral and Brain Sciences, 4, 515-526.

Nishiyori, R. (2016) fNirs:An Emergent Method to Document Functional Cortical Activity during Infant Movements. Front Psychol, 7 (533),

Mar, A.R. (2011) The Neural Bases of Social Cognition and Story Comprehension. Annual Review of Psychology. Annual Review of Psychology, 62, 103-104.

Serpell, J.A. (2003) Anthropomorphism and anthropomorphic selection- Beyond the “cute response” .

Tomasello, M., Kruger, A., & Ratner, H. 1993. 'Cultural Learning.' In: Behavioral and Brain Sciences 16: 495-552

Tomasello, M. 2008. Origins of Human Communication. MIT Press, Cambridge Massachusetts

(18)

Visch, V.T. & Tan, E.S. (2009). Categorizing moving objects into film genres: The effect of animacy attribution, emotional response, and the deviation from non-fiction. Cognition , 110 , 265-272.

Waytz, A., Morewedge, C., Epley N, Monteleone G., Gao, J., Cacioppo J. (2010) Making sense by making sentient: effectance motivation increases anthropomorphism. Journal of

Personality and Social Psychology, 99(3): 410-35.

Watts, G. (2006). Knowing me, knowing ewe. Lancet, 368, 270–272.

Waxman, S. Herrman O., Woodring, J., Medin, D. (2014) Humans (really) are animals: picture-book reading influences five-year-old urban children’s construal of the relation between humans and non-humans animals.

Wheeler, S. C., & Petty, R. E. (2001). The effects of stereotype activation on behavior: A review of possible mechanisms. Psycho-logical Bulletin,127, 797–826

Zimbardo, P. (2008). The lucifer effect: how good people turn evil. Random House Trade Paperbacks.