Making sense of robots Making sense of social robots

6 downloads 301161 Views 826KB Size Report
programming the robot), lead operators to report higher levels of stress, felt responsibility and .... some studies suggesting that participants prefer robots to perform jobs alongside humans (e.g. ...... Geminoid: Teleoperated android of an existing ...
Making sense of robots Running head: Making sense of robots Making sense of social robots: a structural analysis of the layperson’s social representation of robots. / Donner un sens au robot sociaux : une analyse structurelle de la représentation sociale des robots Nuno Piçarra1, Jean-Christophe Giger1,2, Grzegorz Pochwatko3, & Gabriela Gonçalves1,2 1

University of Algarve, Portugal

2

Research Centre for Spatial and Organizational Dynamics - CIEO

3

Institute of Psychology, Polish Academy of Sciences, Poland Authors’ notes

Correspondence to Jean-Christophe Giger, Universidade do Algarve, Faculdade Ciências Humanas e Sociais, Campus de Gambelas, 8005-139, Faro, Portugal. E-Mails: [email protected] (Nuno Piçarra), [email protected] (Jean-Christophe Giger), [email protected] (Grzegorz Pochwatko), [email protected] (Gabriela Gonçalves). This paper is financed by National Funds provided by FCTFoundation for Science and Technology throught project UID/SOC/04020/2013. The study was conducted without any direct or indirect sources of funding from organization or entity with financial interests in robotics. Acknowledgements We would like to thank the two anonymous reviewers for their suggestions and comments. Ref: Piçarra, N., Giger, J.C, Pochwatko, G., Gonçalves, G., (2016). Making sense of social robots: a structural analysis of the layperson's social representation of robot. European Review of Applied Psychology / Revue Européenne de Psychologie Appliquée. doi.org/10.1016/j.erap.2016.07.001.

Making sense of robot Abstract Introduction: Given their novelty, social robots (i.e., robots using natural language, displaying and recognizing emotions) will generate uncertainty among users. Social representations allow making sense of the new, drawing from existing knowledge. Objective: A free association questionnaire was administered to 212 Portuguese adults to identify the social representation of robot. Method: Data was analysed with EVOC 2000 and SIMI 2000 software. Results: The social representation of robot is organized around the ideas of technology, help and future. Differences in the representation according to age, gender and level of education where also identified. Conclusion: The social representation of robot is marked by the conception of it as a tool. This contrasts with the concept of social robots as social agents. Implications for social robot’s acceptance are discussed.

Key-words: social robots; social representation; acceptance of technology; structural analyses.

Making sense of robot Résumé Introduction : Étant donnée leur nouveauté, les robots sociaux (i.e., qui utilisent un langage naturel et montrent et reconnaissent des émotions) vont générer un sentiment d’incertitude chez les futurs utilisateurs. Les représentations sociales permettent de donner du sens aux choses nouvelles à partir des connaissances préexistantes. Objectif : Les participants (N =212 adultes Portugais) participèrent à une tâche d’association livre dans le but d’identifier la représentation sociale du robot. Méthode : Les données ont été traitées à l’aide des programmes EVOC 2000 et SIMI 2000. Résultats: La représentation sociale du robot est organisé autour des idées de la technologie, de l'aide et de l'avenir. Des différences en fonction de l’âge, du genre et du niveau d’éducation sont également identifiées. Conclusion : La représentation sociale du robot est marquée par la conception de celuici comme un outil. Cela contraste avec le concept de robots sociaux en tant qu'agents sociaux. Implications pour l'acceptation des robots sociaux sont discutés.

Mots-clés : robots sociaux; représentation sociale; acceptation de la technologie; analyses structurales.

Making sense of robot Making sense of social robots: a structural analysis of the layperson’s social representation of robots. Donner un sens au robot sociaux : une analyse structurelle de la représentation sociale des robots

1. Introduction The research presented in this paper aims to further the understanding of the layperson’s view of robots and the challenges brought forward by recent developments in the area of robotics. The steep growth curve that characterized robotics over the last decades has brought us close to science fiction reveries like R2-D2 and 3-CPO (Star Wars’ characters). Indeed the concept of an embodied robotic assistant endowed with a social interface (Hegel, Muhl, Wrede, Hielscher-Fastabend, & Sagerer, 2009) has taken form in robots like Snackbot (Lee et al., 2009), Kismet (Turkle, Breazeal, Dasté, & Scassellati, 2006) or Geminoid HI-1 (Nishio, Ishiguro, & Hagita, 2007). Remarkable progresses in artificial intelligence (AI) made possible the development of robots capable of seeing, hearing and communicating in ways similar to humans, i.e. social robots. Capable of using natural language, recognizing and expressing emotional cues, following gaze and gestures, social robots (Fong, Nourbakhsh, & Dautenhahn, 2003) are geared towards the cooperation with humans outside the structured world of the industrial factory. That is, social robots are aimed at tasks usually associated with human expertise. Early examples of this are the three robot wardens tested in a South Korean prison (BBC News, 2011, November 25). These three 150 cm robotic prison guards, are equipped with cameras and sensors and are expected to patrol the wards

Making sense of robot detecting risk behaviors such as violence and suicide among the detainees and thus reducing the workload of the human guards. Although the concept of social robot has gained momentum among researchers and the robotic industry, it is hardly present among lay people. As the recent Eurobarometer report shows (TNS Opinion & Social, 2012), when Europeans were asked to forecast the employment of robots in housework and domestic tasks, 51% answered in 20 or more years’ time and only 4 percent said it was something already common. Results for Portugal follow a similar pattern, with 35% of the participants answering in 20 or more years’ time, while only 8% answered that robots were already used in housework and domestic tasks. Although other studies attempted to understand people’s evaluations of robots (e.g. Piçarra, Giger, Pochwatko, & Gonçalves, 2015), these were based on a pre-set of questions, thus limiting the depth of the knowledge gained. In order to further the understanding of the layperson’s view of this new technology, this paper aims to identify the social representation of robot. Social representations allow people to categorize and make sense of the new, drawing on pre-existing knowledge. They are a guide on how to reason and act towards social objects (Jodelet, 1984, 1989). As such, the identification of the social representation of robot will show the background against which people will make sense of social robots.

2. The disruptive character of robots and the acceptance of innovation models Although industrial robots were initially seen as an increment to the existing industrial machinery, its ability to replace human skills in a series of tasks had a profound effect on job definition and workers’ perception of their role (Shenkar, 1988). Research on the effects of automation showed that the change of focus from mainly

Making sense of robot manual to cognitive tasks (i.e., monitoring the functioning of the machines and programming the robot), lead operators to report higher levels of stress, felt responsibility and reduced interactions with co-workers. The operators also reported that, although the robot eliminated heavy work, the job was now more boring, thus raising the question of worker motivation (Argote, Goodman, & Schkade, 1983). Work displacement, job security, blocked promotions, changes in extrinsic and intrinsic rewards, were other identified effects of automation in industrial settings (Chao & Kozlowski, 1986; Shenkar, 1988). Social robots, on the other hand, are aimed to operate beyond the confines of the industrial plant, representing a departure from current technology, not only in terms of features, but also of interaction possibilities and the place occupied in the social space. Social robots are a discontinuous innovation (Bagozzi & Lee, 1999). As Young et al. (2011) points out, “robot’s social and physical presence, and their tendency to evoke a sense of agency, creates a complex interaction context very different from that of interaction with other technologies and artifacts” (p. 54). Thus, if the industrial robot brought a series of challenges to the organization of labor, the social robot will further those challenges to the organization of social life, blurring the boundaries of what are regarded as exclusively human tasks (see Mutlu & Forlizzi, 2008 for an example of the organizational challenges introduced by the use of a robotic solution). Albeit innovation and technology are frequently equated with progress and welfare, the route is not straightforward (Sabanovic, 2010). As Ram (1987), pointed out, this pro-innovation bias lead researcher to focus on early-adopters (Rogers, 1983), ignoring the common consumer, “…the vast majority of people who have no a priori desire to change may be more typical and even more rational than a small minority of individuals who seek change for its own sake rather than, or in addition to, the intrinsic

Making sense of robot value of the innovations” (Sheth. 1981, p. 274). Contrary to common expectations, what research on innovation has shown is the high degree of uncertainty (Rindova & Petkova, 2007) and anxiety (Fagan, Neill, & Wooldridge, 2003) generated by novelty. As Sheth (1981) pointed out, studying innovation is not only an account of early adopters, but also of those who resist novelty. In order to capture the dynamics of acceptance/resistance of innovation several models and sets of factors have been proposed. For example, Ram & Sheth (1989) identify two types of barriers to innovation acceptance. Functional barriers, like use, value and risk. And psychological barriers, like tradition and image. Other factor like use patterns and information availability have also been associated with the acceptance/resistance of innovation (see Kleijnen, Lee, & Wetzels, 2009 for a review of current models of innovation acceptance). In short, the way people accommodate innovation in their lives is the result of a complex set of representations about usefulness, habit, risk and social norms (Young, Hawkins, & Sharlin, 2009). Social robots, as pointed earlier, represent a clear departure from contemporary conceptions of technology, thus are potential generators of uncertainty and anxiety.

2.1

Review of previous studies The growing presence of robots outside of research and industrial facilities has

drawn the interest of researchers to the general public and casual user’s opinion and perception of, not only the robot per se, but also about what robots could and should do. This question has been tackled using both quantitative and qualitative methods. The following lines draw a brief sketch of the results of this research. Surveys on the opinion about robots show a neutral to positive image of robots (Arras & Cerqui, 2005; Ray, Mondada, & Siegwart, 2008) and technology (Scopelliti,

Making sense of robot Giuliani, & Fornara, 2005). Although older participants considered technology more difficult to use than young and adult participants, they thought it would contribute more to their independence than the other participants (Arras & Cerqui, 2005; Scopelliti, et al., 2005). This positive assistive role of robots is clearly separated from the idea of a robot friend or companion and from the idea that robots can contribute to an increase in personal happiness (Arras & Cerqui, 2005; Dautenhahn et al, 2005). In spite of this positive opinion, a percentage of the participants signaled their discomfort with the prospect of robots performing all the work in the future (Arras & Cerqui, 2005) and associated the idea of robots with job loss, danger, lack of trust, or inhumanity (Ray, Mondada, & Siegwart, 2008). There is a clear preference for the use of robots in practical applications, like household tasks (e.g. vacuum, window cleaning, ironing), while discarding the use of robots in tasks that involve personal relations (e.g. babysitting, company, entertainment, animal care) (Dautenhahn et al, 2005; Ray, Mondada, & Siegwart, 2008; Oestreicher & Eklundh; 2006). Participants also stated preferring a preprogrammed robot, to a more autonomous one (Oestreicher & Eklundh, 2006). It should be noted however that these preferences may be influenced by factors like: the level of expertise and familiarity with robotics or the level of physical disability and perceived health. Ju & Takayama (2011) found that experts and non-experts differ in terms of what they think robots should and should not do, with experts showing a higher confidence in the capability of robots to perform human tasks. Scopelliti et al. (2005) found an expectation gap for non-experts, who though that household tasks like dusting are easier to implement then they are in fact, while at the same time were unaware of robots’ capabilities in areas like home safety control. Oestreicher and Eklundh’s (2006) findings showed that participants with a lesser degree of disability viewed an assistive robot as a doer (performing tasks the

Making sense of robot user does not), while participants with a high degree of disability viewed an assistive robot as a facilitator, allowing them to regain some of the autonomy lost. Cesta et al. (2007) found that elders who perceived themselves as having better health, expressed more positive opinions about the integration of a robot in the domestic environment, showed a more positive emotional response and considered the robot less scary and cumbersome. In terms of of human-robot collaboration results are inconclusive, with some studies suggesting that participants prefer robots to perform jobs alongside humans (e.g. Takayama, Ju, & Nass, 2008), while other studies suggest the opposite (e.g. Ju & Takayama, 2011). Regarding robot appearance, the majority of the participants thought of robots as machine-like. The domestic/assistive robot should be a small machine, able to communicate in a human like fashion, but should not have a zoomorphic or humanoid form (Arras & Cerqui, 2005; Ray, Dautenhahn et al, 2005; Ray, Mondada & Siegwart, 2008). Appearance may play an important role in the perception of the robot’s capabilities and suitability for certain tasks. Participants presented with images of robots attributed significantly more communication and emotion capabilities to human-like and animal-like robots than to machine-like robots, but the same level of mobility and information processing power (Lee, Lau, & Hong, 2011). Kamide, Kawabe, Shigemi, and Arai (2013) analyzed the ratings of three humanoid robots and found differences for gender and age group. Female participants rated robots higher on familiarity than male. Older male and female participants rated robots higher in humanness. Middle-age and old-age male participants give higher ratings to humanoids than young-age and adolescents. Female participants in the middle-age group rated utility of humanoids as higher.

Making sense of robot Although the research results presented above provide already a large set of very useful answers, little is still known about the socio-cognitive processes underlying the laypersons understanding and expectations of what robots are, where they should operate and their role in society. When confronted with uncertainty people tend to draw on what is familiar in order to make sense of their surroundings. As such, it is expectable that people will build their opinions and knowledge about social robots based on what they already know about current technologies. In order to explore this dynamic processes, this paper proposes the use of the framework afforded by the social representations theory. The following lines provide a brief sketch of the theory.

3. Defining Social Representation When Serge Moscovici, in the early sixties, set forward the theory of social representations, he wanted to tackle the question of how lay people in modern, ever changing, societies accommodated the growing body of technical and scientific knowledge placed at their disposal by the social media (radio shows, television and newspapers), and turned it into something useful for their daily life. In other words, he wanted to know how common knowledge is formed, how individuals in social interaction, build theories about social objects and thus make possible communication and the organization of behavior (Vala, 1993). The social representation is a form of knowledge that aims to transform what is strange into something familiar, by anchoring novelty to already existing and stable knowledge structures (Moscovici, 1961). Social representations are a form of shared knowledge, allowing social actors (individuals and communities) to frame and act on social objects. By presenting norms and prescriptions, they help making sense of daily experience. They are a cultural, linguistic and communicational phenomenon tightly connected to social structures.

Making sense of robot Their content and internal organization supposes a subject, an object and a process of construction, expression, interpretation and symbolization. Social representations are a practical and instrumental form of knowledge (Vala, 1993). In a certain sense, social representations are a guide for action, since they include a desirable course of action, they give meaning to the social object and its context and they give meaning to behavior itself (Vala, 1993).

3.1

Dialectical dynamics: Anchoring and objectification Social representations integrate a cognitive and a social component. The

cognitive component which is the individual’s role in actively appropriating and restructuring reality, with the aim of anchoring and stabilizing social objects. The social component, which is the product of interactions, the social production of a common reality by the group, with the aim of creating collective objects and equilibrium (Abric, 1996). This appropriation is done by a dialectical process of anchoring and objectification. Objectification is the process of turning an abstract knowledge into something concrete. The available information is de-contextualized, selected, simplified and organized in what is to be a new fact, turning it into a frame for categorizing and interpreting new information. This new frame of categorization is then given correspondence to a natural reality. Anchoring works on one hand as a starting point to think new objects, and on the other hand as the representation of the new object that emerges from the process of objectification. Anchoring is the process of integrating new information on the already existing system of categories and relations. As such, it will modulate what and how, new information will be integrated in the already existing network of meanings. “It generates a system of interpretation, it offers a framework for

Making sense of robot the determination of behaviors in creating expectations, needs and anticipations” (Abric, 1996, p. 78).

3.2 The structural approach According to Abric’s (1993) theory of the central nucleus, a social representation is structurally composed of a central nucleus (or central core) and a peripheral system. The central nucleus is made of one (or several) elements of the representation, and is characterized by having the generative function of ascribing meaning and organizing the elements of the representation (Guimelli, 1993). The central nucleus is directly linked and determined by historical and social conditions, being strongly marked by the group’s system of norms. It is consensual and collectively shared by a social group. It is stable, coherent and resistant to change. It gives the representation a sense of continuity and consistency. It is in a certain way independent from the immediate social context (Abric, 1993). This central core structures the meaning of the whole representation, including the peripheral elements. It is a necessary condition for the representation’s role as a meaning making tool. From a behavioral standpoint, the central nucleus plays a central role in the organization of values, attitudes and actions (Abric, 2003). While the central core is normative, the peripheral system is functional, grounding the representation on reality. “It is the peripheral elements which can withstand the variations between individuals, between subgroups, and over time” (Flament, 1994, p.7). The peripheral system’s role is turning the central core norms and prescriptions into concrete courses of action, answering to concrete daily challenges. To do so the peripheral system is sensitive and determined by context, showing flexibility and accepting contradictions. Given this characteristics, the peripheral system serves a

Making sense of robot regulatory function in the adaptation of central core norms to new situations. It functions as a buffer, absorbing new information and events that challenge the core prescriptions of the representation, serving as a protection mechanism. This flexibility also allows for individual differences and creativity, integrating personal experiences and history in individualized social representations, but keeping them organized around the central core shared by the social group. From the interplay of this dual structure, central core and peripheral system, emerges the social representation apparently incongruent character, at once stable and fluid, consensual but marked by interindividual differences. The presence of conflicting ideas and practices can result in several outcomes for the social representation (Abric, 1993): if the situation is perceived as reversible, the transformation is limited to the peripheral elements of the representation (the core remains unchanged) and limited in time. If the situation is perceived as irreversible, the results can be, a resistance to transformation, through rationalization and latter explanation of the facts; a gradual transformation, if the new facts are not in total disagreement with the central nucleus norms; or a total rupture, questioning the prescriptions of the central nucleus, leading to its change and the emergence of a new social representation, in line with the new facts.

3.3

Social representations and behavior Social representations play a role in the creation of expectations and needs

(Abric, 1996), in communicating and organizing behaviors (Vala, 1993). As such, it is reasonable to expect them to play a role in individual behaviors. This hypothesis has received empirical support. Guimelli (1993), for example, reports the effect of individual representations, in nursing student’s choice of wanting to work in the public or private health systems. After looking at the students’ representation of the nurse’s

Making sense of robot job, it become clear that their choice of future work place, was the result of the students’ perception of which of those two options, public or private sector, allowed them better to fulfill what they though (social representation) where the nurse’s function. More recently, Gomes, & Nunes (2011) argued for a relation between the social representation of sex, as portrayed in newspapers and magazines, and an inconsistent use of condoms. Apostolidis & Dany (2012) studied how the social representation of risk informed health practices regarding AIDS and psychoactive substance use. Research also suggests that the social representation operates differently according to the relation the person entails with the social object. Moliner and Gutermann (2004) found that depending on the kind of relation that the person established with the social object (i.e., deviating people), the social representation would take a descriptive role (when participants had little to no contact with the social object), or an explanatory role (when participants had frequent contact with the social object). Salès-Wuillemin et al. (2011) studied how the social representation of hospital hygiene changed from student nurses to professional nurses, showing how professional training and common knowledge intertwine. In a similar vein, Gangl, Kastlunger, Kirchler, & Voracek (2012), compared experts and laypeople’s social representation of the financial and economic crisis. Lin, He, Jin, Tao, and Jiang (2013), on the other hand focused on gender and health, identifying differences in the social representation of pain, which lead to differential social expressions.

4. Objectives and overview of the research This study aims to determine the social representation of robot in a sample of Portuguese adults in order to understand how the concept of social robot will fit it (or misfit). Following a structural approach, a four quadrant diagram was constructed to

Making sense of robot identify the candidates to the central nucleus and the periphery of the social representation. A similitude analyses was conducted to identify the relations between the elements of the social representation. The social representation of robot was further analyzed by gender, age and years of schooling.

5. Method 5.1 Participants The convenience sample is composed of 212 Portuguese participants (128 women and 76 men; 8 not reported). Data was collected in the University of the Algarve, Gambelas campus, and at an adult education center in the district of Faro. Table 1 shows the socio-demographic characteristics of the participants. 5.2 Material and procedure The method used to collect data was free evocation (Rouquette & Rateau, 1998), with participants receiving the following instructions: “Please write the ideas (names, adjectives...) that pop up into your mind when you listen to the word robot. Use a line for each idea.” No limit number of ideas was given. Participants where only asked to write their evocations, as this kept the task simpler, allowing to circumvent the logistic constraints of collecting data in different locations. The instructions were accompanied by an explanation of the voluntary character of the participation, of the confidentiality of the data and the explicit statement that they could stop to respond whenever they wanted if they felt uncomfortable with the task.

6. Results 6.1 Coding

Making sense of robot All data was transcribed to a spreadsheet in order to be prepared for lexicographical analysis. Since some responses were given in the form of a sentence, they were replaced by a word that summarized the idea. If the sentence encompassed several ideas, several words would be used in order to represent each of the ideas.

6.2 Lexicographical analyses of the social representation of robot The lexicon for this study is composed of a total of 1666 words, with 581 unique occurrences. It was built following Vergès, Scano, and Junique (2002) recommendations. On average, participants evoked 7.74 words (SD = 4.63). The number of evocations varied between 1 and 22. Evocations were organized by frequency and evocation order, in a four quadrant diagram, which allows the identification of what ideas are central to the social representation and what ideas compose the peripheral system (Vergès, 1992; Abric 2003; for a review of other methods see Moliner & Guimelli, 2015). In order to conduct a lexicographical analyses three values must be determined: mean frequency, minimum frequency and mean order. Mean frequency and minimum frequency were calculated through the analysis of the frequency distribution of the evocations (see Vergès et al., 2002). The frequency table allows the identification of three distribution zones:  Many words and very low evocation frequency (e.g. 419 words are present only 1 time in the lexicon);  Few words and low evocation frequency (e.g. 4 words are present 8 times in the lexicon);  Few words and high evocation frequency (e.g. 1 word is present 26 times in the lexicon).

Making sense of robot Minimum frequency was considered 6, which represents the point where word frequency changes from many words and very low evocation to few words and low evocation. Mean frequency was considered 13, which represents the point where word frequency changes from few words and low evocation to few words and high evocation frequency. Literature on social representations (e.g. Dany, Urdapilleta & Lo Monaco, 2015) suggest the use of 2.5 as the mean order. The data was analyzed using the software EVOC 2000 (Vergès et al., 2002). Table 2 shows the words frequency distribution and table 3 shows the four quadrant diagram representing frequency and order of evocation. 6.2.1 Analysis of the candidates to the central nucleus According to Vergès, Tyszka, and Vergès (1994), the elements of the central nucleus display two features: consensuality and easiness of recall. That is, if people are asked what their ideas about robots are, the ideas pertaining to the central nucleus would be those with a recollection rate above the average of the ideas recalled (consensus) and a recollection order below the average recollection order of the ideas recalled (evocation readiness). In the four quadrant diagram the ideas more likely to pertain to the central nucleus are represented in the superior left quadrant (see Table 3). In this quadrant, machine is the idea with the highest frequency (100) and lowest rank order (1.9). That is, when participants are asked about the word robot, the idea of machine is not only the most evoked, but is also the one invoked first more frequently. The idea of machine is accompanied by the idea of automatic. It should be noted however that despite being among the easily evoked ideas (order = 2.4), automatic may not be a very consensual idea given its evocation frequency (19).

Making sense of robot In brief, the word robot evokes an idea of a machine that performs by itself, that is, automatic.

6.2.2 Analysis of the of the candidates to the first periphery. The upper right quadrant of the diagram is called the first periphery (Vergès, 1992; Abric 2003). The words presented here, although having a frequency above the mean frequency, are ranked below the mean order. Even though these elements are peripheral in the representation, they keep a close connection to the central nucleus, and function as a buffer to external threats (Abric, 1993). That is, they serve a regulatory function, adapting central core norms not only to daily but also new situations, absorbing new information and events even when they contradict the consensual nucleus elements. The ideas identified in the first periphery, present a more concrete vision of what a robot (the machine) can be. A quick read through the list of ideas evoked shows words related to the material used to build it, capabilities, usefulness and socio-economic impact. The upper right quadrant depicts robots as: a technology, derived from artificial intelligence, electronics, programming and computers. The work of science and innovation. An evolution, the future. Something that will help, facilitate chores and replace men in some tasks. Industrial robots, domestic robots, entertainment robots (puppets) are examples. Despite this futuristic view, the idea of robot still evokes images somewhat related to the industrial age like mechanization, metal and mechanical. The socio-economic effects of the deployment of robots is also present in the evoked ideas, namely through the idea of unemployment. However, it is not clear how

Making sense of robot this is related to ideas like replacement of men, help or mechanization, since these ideas seem to be equated with replacement of men in unpleasant or dangerous tasks. Interestingly, the concept of robot seems to evoke the age old dichotomy between reason (intelligent) and emotion (without feelings), thus underlining one of the tenets of the social representation theory, that is, how the old is used to make sense of the new (Moscovici, 1961). This however, contrasts with the current trend in technology towards the humanization of interfaces. Finally, the significant contribution of popular culture for the construction of social representations is also present through the association of the idea of robot with that of movies and robocop. This suggests the major role popular culture plays, molding promises and perils people associate with a future populated by robots. Curiously the pop culture icon more frequently evoked is robocop, a cyborg police agent that is in clear contrast with the idea of the robot machine present in the first quadrant. By definition, the first periphery plays an important role both in maintaining the nucleus stability and in accommodating new and unfamiliar objects. This is visible in the diverse and contrasting set of ideas presented here, where ideas like artificial intelligence and mechanical are set side - by – side. If on the one hand the general idea of machine is prevalent in the evoked ideas, on the other hand the popular icon most evoked is that of a cyborg, part men, part machine, torn apart by his ambiguous nature. In short, the idea of machine accommodates to a wide range of meanings and embodiments. In spite of the presence of the idea of unemployment, the evoked ideas present a positive view of robots as helpers, replacing men in hard tasks, providing an evolution relative to prior technologies.

Making sense of robot 6.2.3 Analysis of the candidates to the contrast zone and the second periphery. The lower left quadrant of the diagram is called the contrast zone (Vergès, 1992; Abric 2003). The words present in this quadrant have an order of evocation above the mean order of evocation, but their evocation frequency is lower than the mean frequency of evocation. The contrast zone sometimes reveals complementary ideas or the presence of a subgroup with a different social representation (Vergès, 1992; Abric 2003). This research did not identify any ideas pertaining to this quadrant. This absence reinforces the consensual character of the ideas present in the first quadrant (upper left quadrant). The lower right of the diagram is composed of words with a frequency and an evocation order bellow the mean evocation and mean order. These are the more peripheral elements of the representation and they constitute the second periphery (Vergès, 1992; Abric 2003). Among the ideas found in the lower right quadrant there are references to the technological aspects of robots (e.g. robotics, electrical, artificial, autonomous), its social impact (e.g. useful, development, improvement) and its ludic character (e.g. entertainment, toy, fiction, I robot, star wars). Despite their low frequency and recall order, it is interesting to note that these ideas are still within the framework provided by the first periphery, representing cases of more general ideas or concepts.

6.2.4 Synthesis of the representation. Analysis of the four quadrant diagram shows the idea of machine as a strong candidate to the central nucleus of the social representation. Although the idea of automatic is also present in the upper left quadrant, it presents a lower frequency than

Making sense of robot some words of the first periphery. That is, although the idea of automatic is easy to recall, ideas like technology or future are more frequently recalled. The ideas pertaining to the first periphery can be roughly organized around robot characteristics (e.g. technology, metal, artificial intelligence, electronics), social consequences (e.g. help, facilitates, replaces men, unemployment) and time of deployment (future). Like it is proposed by the structural model of social representations (Abric, 1993; Moliner & Guimelli, 2015), the first periphery proposes concrete embodiments for the concept of robot, while allowing the co-occurrence of contradictory representations like the high tech machine (e.g. artificial intelligence, programming, innovation) vs. the old industrial machine (e.g. mechanization, mechanical, metal), or the helpful machine (e.g. help, facilitates, replaces men, evolution) vs. the disruptive machine (e.g. unemployment). In short, if on one hand, the idea of robot as a machine is consensual, on the other, there is plenty of room for diversity in terms of embodiment and competences expected, with the co-occurrence of several robot “models” (e.g. industrial robot, domestic robot, robocop). However, it should be noted that that this diversity occurs within the framework of machine. A machine which is intelligent but without feelings. This is at odds with the current trend toward social interfaces and social robots.

6.3 Similitude analysis of the social representation of robot. Central to the structural approach to social representations is the notion that meaning derives, not from its elements alone, but from their organization. As such, besides identifying what are the composing elements of the representation, a second step is necessary, studying how these elements are interconnected and what meaning emerges from these relations (Rouquette & Rateau, 1998). This can be accomplish

Making sense of robot using the similitude analysis. This is a technique derived from graph theory (see Degenne & Vergès, 1973 for a description), which allows the study of the interrelations of the elements composing a social representation. With it, is possible to display graphically the organization of these elements, in what is called the maximum tree (i.e., l’arbre maximum; Degenne & Vergès, 1973). In this representation the vertices are occupied by the words pertaining to the representation. These vertices are connected by edges that indicate the degree of connection between these words. This allows seeing which ideas have more connections, how strong they are and if words connect in such a way as to give rise to new ideas (Degenne & Vergès, 1973), thus providing a more dynamic view the social representation’s structure. For the purpose of this study the ideas identified as candidates to the central nucleus and the first periphery where organized into 25 categories and then analyzed with the software SIMI 2000 (Junique, Barbry, Scano, Zeliger, & Vergès, 2002). Figure 1 shows the results. The ticker the line, the stronger the relation between the ideas (see Vergès, 2001). The representation of robot is organized around the nodes of technology and future. The idea of technology is connected to science, machine, artificial intelligence, help and innovation. The idea of machine is connected to that of computer. The idea of help is connected to facilitate and replaces men. Although the idea of machine is still present, it lost the centrality it had in the analysis of the quadrants. The similitude analysis of the social representation of robot portrays it as a technology which can help humans. Also significant is the strong relation between facilitates and replaces men. The second organizational node of the tree is the idea of future. This idea is connected to movies, evolution, electronics and industrial robot. Although given concrete uses

Making sense of robot (e.g., industrial, electronics or movies) the robot is projected as something belonging to the future. In short, robots are seen as the helping technology of the future.

6.3.1 Comparison by socio-demographic characteristics. One of the characteristics of social representations is their ability to form a coherent whole, while allowing the development of more individual and contextual representations. In order to better understand these dynamics, the sample was split by gender, age (two groups formed using the median, 32 years), and years of school (two groups, up to 12 years of schooling and university degree or frequency). The analysis used the same 25 categories as above.

6.3.1.1 Gender. Figures 2 and 3 show the organization of the representation for female and male participants. The social representation of robot in female participants (n = 128) is organized around the nodes of technology and help (see Figure 2). Technology is connected to innovation, machine, science, evolution and future. Also noteworthy is the strong relation between future and replaces men. Although the role of technology is similar to what was found for the total sample, the idea of help assumes a more central role in the organization of the representation. This node presents two distinct views of robot, the domestic robot and the puppet. The male participant (n = 76) social representation is organized around two nodes, artificial intelligence and help (see Figure 3). Artificial intelligence is connected to programming, metal, industrial robot, mechanical and future. The node of help is connected to movies, machine, replaces men and unemployment. It is interesting to note

Making sense of robot that concrete aspects of what is a robot (e.g. artificial intelligence) assume a more central role in the organization of the male’s social representation than in female’s representation. Also, unlike the female social representation, the model of robot in the male representation is the industrial robot. Also noteworthy is the connection between technology, unemployment and help, which suggests that for the male participants the use of robots might have some negative social effects. This connection is not present in the female representation. The representation is gendered in the way that it reflects the traditional gender roles, patriarchal norms and family dynamics. Robots are for men a form of intelligence to be controlled and mastered (programming) associated with work, while they are for women technological tools to be used in a domestic context, and a technology also seen in movies. Such results are in line with previous research on gender and technology. For example, it was shown that males considered computers as an artefact to be mastered whereas females considered computers as an instrumental tool to complete a task (Morritt, 1997; Turkle, 1988). In other words, the social representation of robots is associated to the classical patriarchal dichotomies in terms of economic roles (male breadwinner vs. female homemaker), space (outside home / public sphere for men vs. inside home/private sphere for women), agency (action, i.e., mastering for men, vs. passivity, i.e., passive use for women) and masculine vs. feminine culture.

6.3.1.2 Comparison by age. For the comparison by age, the group was divided using the median (32 years), resulting in two groups with 98 participants each. The social representation of robot, for the group aged bellow 32 years, is organized around the idea of technology (see Figure 4). Noteworthy is the diversity of ideas connected to technology, ranging from artificial

Making sense of robot intelligence to movies. Once again the representation includes the idea of an intelligent machine, an industrial robot, that helps and replaces men in the future. For the group aged above 32 years, the representation of the robot is organized around the nodes of help and future (see Figure 5). Help is connected with technology and unemployment, while future is connected with computer, replaces men and evolution. Albeit the idea of a technology that replaces men in some tasks is well received by both groups (facilitates, evolution), it raises some concerns among the older participants (unemployment). It is interesting to note that, while younger participants equate the idea of robot with artificial intelligence, older participants view it as something akin to computer. Another characteristic of the subgroup, age above 32 years, is that robots portrayed in movies or toys (puppets) are considered apart from those of real life.

6.3.2 Comparison by years of schooling For the comparison by years of schooling, the group was divided in two groups, up to 12 years of schooling (n = 90) and university degree (n = 116). For the first group, the social representation is organized around the node of help (see Figure 6). Help is connected with intelligent, future, technology and puppet. Two main ideas of robot can be identified, the industrial robot and the robots from science fiction movies. The representation for the subgroup university degree, is organized around the node technology, with artificial intelligence forming a second node. (see Figure 7). The idea of help is also present, but is not central. The representation for this group is more complex, integrating diverse ideas like artificial intelligence, mechanical, innovation and science. This means that this subgroup has more ideas available to think about the

Making sense of robot effects robots will have in their lives. Also noteworthy is the idea that robots are without feelings. This idea is connected with that of machine and metal. It is not clear the ontological significance of this attribution. Nonetheless, both a lack of feelings derived from the robot’s mechanical nature, or its opposition to what is considered to an inherently human trait, will have an effect on the acceptance of robots in general, and social robots in particular. 6.4 Synthesis of the similitude analysis. The similitude analysis provides a second step in the study of social representations, allowing the study of the interrelations of its elements. From this analysis surfaced a rearrangement of the elements previously identified as candidates to the central nucleus and the first periphery. The social representation of robot for the total sample is organized around the nodes technology and future, with the idea of machine losing the centrality previously identified. In general, the robot is viewed as a technology, a machine (computer) with artificial intelligence, an innovation brought by science. It helps and facilitates work, replacing men. Robots are seen as a future event, an evolution of electronics, that will take the form of industrial robots. This future with robots can already be seen in some movies. Finally, robots are viewed as competent but emotionless machines. The themes of technology, help and future are common to the various subgroups studied, underlining the homogeneity of the representation. Despite these common elements, the similitude analysis displayed different structural organizations for the social representation regarding gender, age and schooling years. A brief description of these differences follows.

Making sense of robot The male representation is organized around a concrete characteristic of robot (artificial intelligence), while the female representation is organized around a vaguer idea (technology). They also privilege different uses for robots, industrial in male’s representation, domestic in female’s representation. Although both representations ascribe a central role to help, males show some apparent concern with unemployment. Participants aged bellow 32, organize their representation around the idea of technology, while those aged above 32, organize it around the ideas of help and future. Participants aged above 32, also show some concern with unemployment. Besides organizing their social representations around different nodes (help for participants up to 12 years, technology for participants with a university degree), these two subgroups also present a different level organizational complexity. It should be noted however, that these differences represent not diverse conceptions of what robots are, but a focus of the participants on particular aspects of the social representation which are more relevant for their social contexts.

7. Discussion The results presented above are in line with previous research. In 1983, Argot et al., interviewed the workers of a plant during the installation of a robotic unit. When prompted with the open question “How would you describe a robot to a friend?” the answers were: mechanical man, pre-programmed machine, something that loads machines, increases productivity or reduces manual work. More recently, the results of the special eurobarometer 382, titled “Public Attitudes Towards Robots” (TNS Opinion & Social, 2012), a report that describes EU residents’ general attitude towards scientific discoveries, technology and robotics, presented a similar panorama. Participants were shown a picture of an industrial robot (an automated programmable arm filling boxes)

Making sense of robot and a picture of a humanized home helping robot. They were then asked to rate how much, each of the pictures fitted their image of a robot. Around 80% of the participants stated that the image of the industrial robot fitted well with the image they had of robots, while 66% of the participants stated that the image of the humanized home helping robot fitted well with the image they had of robots. This suggests that the image of the industrial robot, automated, programmable, mechanical arm is still very pervasive amongst the layperson. In the case of Portugal, the gap was smaller with 64% of the participants stating that the industrial robot fitted well their image of robot and 55% stating that the humanized home helping robot fitted well their image of robot. In short, the present results confirm that, in Portugal, the idea of the robot as an industrial, mechanical, technology, is central. Since previous reports did not find significant differences among European’s view of robots and technology (TNS Opinion & Social, 2012), the results of this research may provide a good starting point for a cross cultural exploration of the social representation of robot. It is also noteworthy to point that, in more than 30 years, in spite of all the technological developments, very little seems to have changed in the way people perceive robots. In spite of this apparent inertness, the dynamic process of objectification and anchoring is visible at work in the representation of robot. Indeed, elements pertaining to diverse technologies (e.g. computer) and contexts (e.g. industrial processes) are integrated in the representation of robot. Drawing from a set of characteristics common to traditional machinery (e.g. metal, mechanical, fast) and current technology (e.g. electronics, artificial intelligence) people built a shared representation of what a robot is. Drawing on their conception of work, people built a shared representation of what it does (help, facilitates, replaces men), and where it operates (industrial and domestic

Making sense of robot robots). Also, the role of cultural icons portrayed by science fiction movies should not be overlooked in this process. Although it is not possible to anticipate what elements, if any, of the social representation of robot will be used in the construction of the socially shared knowledge about social robots, these elements will surely be anchored in daily practices. Indeed, the social representation of robots is gendered and rooted in gendered practices. Men and women’s everyday life practices (e.g., at work, at home) are guided and controlled by social routines, schemas, conventions, norms, and ideologies associated to gender that prescribe how men and women should make sense, experience and use technology. Actually, the present study showed that men associated robots to workplace while women associated them with domestic contexts. Moreover, although both genders perceived robots as a helping technology, men perceived robots as a threat (i.e., replacing men and creating unemployment) while women associated robots with help in a domestic context. Such results might indicate that women may accept more easily social robots than men, and the laundry might be the first entry to social robots in our daily life. First, women are still mainly responsible for the house related tasks, and the introduction at home of social robots could be seen as way to free women from domestic drudgery. Previous research has showed that “compared with men, the women talked more explicitly about the importance of domestic technologies in their lives” and “described how technologies helped them – with their chores, with childcare.” (Levingston, 1992, p. 5). Moreover, Carpenter Davis Erwin-Stewart Lee, Bransford and Vye (2009) showed that after viewing video clips presenting two exemplars of social robots that could be use at home, participants reported that they could mainly be used for chores like washing dishes, doing laundry, general cleaning, ironing clothes. Second, women seem to use technology like gaming, computers and social networks for

Making sense of robot social interaction more than men (e.g., Veltri, Krasnova, Baumann, & Kalayamthanam, 2014). The fact that social robots are programmed to communicate on a human mode may facilitate their acceptance. For example, Carpenter et al. (2009) showed that participants reported that socialness of social robots (e.g., speech) would made easier their use. However, a new technology as social robots could be threatening and unfamiliar, especially when it challenges a fundamental dimension of gender identity and roles. Indeed, “technologies are represented as objects of identity projects – objects that may stabilize or de-stabilize hegemonic representations of gender." (Oudshoorn & Pinch 2003, p. 10). For example, Carpenter et al. (2009) reported that participants did not want social robots to be used for childcare or companionship. To sum up, first, given its role as organizers of knowledge, the study of social representation uncovers not only the tendency towards robots, but also what robots are, where they are and why they are there. Participants in this study portray the robot as a technology, a sort of machine with an artificial intelligence, an innovation produced by science (what), deployed on industrial or domestic settings (where), performing hard, dangerous and mundane tasks, helping, assisting and replacing men (why). Given this, it can be said that there is a fairly positive social representation of robots, as they are equated with technological progress and the pursuit of a “better life”. These results are also in line with the results of previous research, including that conducted more than 30 years ago. Second, although the social representation of robot is quite consensual in terms of its components, there are clear organizational differences within the subsamples studied. One example is the connection between the ideas of replacement of men, help and facilitates. If on the one hand, these are common themes for all subsamples suggesting a positive view, on the other hand, in the case of male participants and

Making sense of robot participants above 32 years, the ideas of technology and help are connected with the idea of unemployment, which is per se a negative outcome. That is, if for participants in general the robot is a welcomed help, for some participants it may be a source of anxiety. Another example, is the subsample of participants with university degree, who describe the robot as a machine without feelings. Thus, for these participants the notion of a social robot may stir some resistance. Both examples are in line with the results reported by Ray, Mondada, and Siegwart (2008) that showed that, in spite of a general positive view of robots, some participants showed concern with job loss, trust and inhumanity (absence of affection). This underlines the need for further studies in this area. Finally, if the layperson idea of robot did not change much in the last 30 years, the same cannot be said about the robotic industry. This means there is a gap between the layperson expectations and current trends in robotics research. While people still think of a “high-tech” mechanical tool, something that will facilitate, or replace them in unpleasant or dangerous tasks, the industry is preparing the social robot. A robot that recognizes and expresses emotions, uses natural speech and engages in social interactions. A robot that will take the role of an “high-tech” co-worker, a teacher, a nurse, a companion. This expectation gap was also identified by previous research, people expect from robots the performance of practical daily tasks and a limited autonomy. And although people mention preferring to communicate in a natural fashion, i.e. using verbal instructions, they imagine future robots as small machines that can be easily fitted in the house (Dautenhahn et al, 2005; Ray, Mondada, & Siegwart, 2008; Oestreicher & Eklundh; 2006). The acceptance of social robots will depend, not only on the efforts to narrow this expectation gap, but also, on the understanding of how people make sense of what is the meaning of work, help and replacement.

Making sense of robot

8. Further research Although the similitude analysis allows the study of the interrelations between the elements of the social representation, further studies are necessary to confirm the structure of the social representation and the central nucleus (see Moliner & Guimelli, 2015 for a review of methods). The use of other evocation stimulus (e.g. different types of robot and technologies), requesting evocations about the use of robots in different professional and social settings (e.g. factory, warehouse, hospital, school), or further exploring the social representation of different social groups (e.g. unemployed, retired), and nationalities, could bring further insights on the social representation dynamics and structural organization. The use of a longer timeframe could bring some insights on the changes brought by the increasing presence of technology and robots, to the social representation’s structural organization, namely the buffer role of the first periphery. It could also allow the application of a developmental outlook on the presence and acceptance of robots.

Conflicts of interest statement The study was conducted without any direct or indirect sources of funding from organization or entity with financial interests in robotics.

Making sense of robot References Abric, J.-C. (1993). Central system peripheral system: their functions and roles in the dynamics of social representations. Papers on Social Representations - Textes sur les Représentations Sociales, 2, 75-78. Retrieved from http://psych1.lse.ac.uk/psr/PSR1993/2_1993Abric.pdf Abric, J.-C. (1996). Specific processes of social representations. Papers on social represenations - Textes sur les Représentations Sociales, 5, 77-80. Retrieved from http://psych1.lse.ac.uk/psr/PSR1996/5_1996Abric.pdf Abric, J.-C. (2003). La recherche du noyau central et de la zone muette des représentations sociales. In: Abric, J.-C. (ed.) Méthodes d’études des représentations sociales [Methods for studying Social Representations], pp. 59–80. Ramonville Saint Agne, Erès. Apostolidis T., & Dany, L. (2012). Pensée sociale et risques dans le domaine de la santé : le regard des représentations sociales [Social thought and risks in the health field: The social representations perspective]. Psychologie Française, 57, 67-81. doi: 10.1016/j.psfr.2012.03.003 Argote, L., Goodman, P., & Schkade, D. (1983). The Human Side of Robotics: How Worker's React to a Robot. Sloan Management Review, 24 (3), 31-41. Arras, K. & Cerqui, D. (2005). Do we want to share our lives and bodies with robots? A 2000-people survey (Technical Report Nr. 0605-001). Autonomous Systems Lab, Swiss Federal Institute of Technology, EPFL. Retrieved from http://infoscience.epfl.ch/record/97585/files/SurveyPaperArrasCerqui.pdf Bagozzi, R. & Lee, K.-H. (1999). Consumer resistance to, and acceptance of, innovations. Advances in Consumer Research, 26, 218 – 225. Retrieved from http://www.acrwebsite.org/search/view-conference-proceedings.aspx?Id=7902 BBC News (2011, November 25). Robotic prison wardens to patrol South Korean prison. BBC News. Retrieved from http://www.bbc.co.uk/news/ technology15893772 Carpenter, J., Davis, J. M., Erwin-Stewart, N., Lee, R. L, Bransford, J. D., & Vye, N. (2009). Gender Representation and Humanoid Robots Designed for Domestic Use. International Journal of Social Robotics, 1, 261–265. doi 10.1007/s12369-0090016-4 Cesta, A., Cortellessa, G., Giuliani, M., Pecora, F., Scopelliti, M., & Tiberio, L. (2007). Psychological Implications of Domestic Assistive Technology for the Elderly. PsychNology Journal, 5 (3), 229 – 252. Retrieved from http://psychnology.org/File/PNJ5(3)/PSYCHNOLOGY_JOURNAL_5_3_FULL.pd f#page=9 Chao, G. & Kozlowski, S. (1986). Employee Perceptions on the Implementation of Robotic Manufacturing Technology. Journal of Applied Psychology, 71 (1), 70-76.

Making sense of robot Dany, L., Urdapilleta, I., & Lo Monaco, G. (2015). Free associations and social representations: some reflections on rank-frequency and importance-frequency methods. Quality & Quantity, 49, 489-507. doi: 10.1007/s11135-014-0005-z Dautenhahn, K., Woods, S., Kaouri, C., Walters, M., Koay, K., & Werry, I. (2005). What is a robot companion - friend, assistant or butler?. IROS ’05, International Conference on Intelligent Robots and Systems, 1192 - 1197. doi : 10.1109/IROS.2005.1545189 Degenne, A., & Vergès, P. (1973). Introduction à l'analyse de similitude [Introduction to similitude analysis]. Revue française de sociologie, 14 (4), 471-511. Fagan, M., Neill, S., & Wooldridge, B. (2003). An empirical investigation into the relationship between computer self-efficacy, anxiety, experience, support and usage. Journal of Computer Information Systems, 44, 95 – 104. Retrieved from http://digitalcommons.calpoly.edu/cgi/viewcontent.cgi?article=1027&context=mkt _fac Flament, C. (1994). Consensus, salience and necessity in social Representations technical note. Papers on Social Representations - Textes sur les Représentations Sociales, 3, 1-9. Retrieved from http://www.psych.lse.ac.uk/psr/ Fong, T., Nourbakhsh, I., & Dautenhahn, K. (2003). A survey of socially interactive robots. Robotics and Autonomous Systems, 42 (3-4), 143-166. doi: 10.1016/s09218890(02)00372-x Gangl, K., Kastlunger, B., Kirchler, E., & Voracek, M. (2012). Confidence in the economy in times of crisis: Social representations of experts and laypeople. The Journal of Socio-Economics, 41(5), 603– 614. doi:10.1016/j.socec.2012.05.018 Gomes, A. & Nunes, C. (2011). Representação Social do Sexo nas Revistas e Jornais Portugueses. [Social representation of sex in portuguese newspappers and magazines]. Revista Interamericana de Psicología/Interamerican Journal of Psychology, 45 (1), 11-19. Retrieved from http://www.redalyc.org/articulo.oa?id=28421134003 Guimelli, C. (1993). Locating the central core of social representations: towards a method. European Journal of Social Psychology, 23 (5), 555 – 559. doi: 10.1002/ejsp.2420230511 Hegel, F., Muhl, C., Wrede, B., Hielscher-Fastabend, M., & Sagerer, G. (2009). Understanding Social Robots. Second International Conferences on Advances in Computer-Human Interactions, ACHI ’09, 169-174. doi: 10.1109/ACHI.2009.51 Jodelet, D. (1984). Représentation sociale: phénomènes, concept et théorie [Social representation: phenomena, concept and theory] . In S. Moscovici. Psychologie sociale (pp. 357-378). Paris, PUF. Jodelet, D. (1989). Représentation sociale: un domaine en expansion [Social representations: an expanding domain]. In D. Jodelet. Les Représentation sociales (pp. 31-61). Paris, PUF. Ju, W., & Takayama, L. (2011). Should Robots or People Do These Jobs? A Survey of Robotics Experts and Non-Experts About Which Jobs Robots Should Do. IROS

Making sense of robot ’11, International Conference on Intelligent Robots and Systems, 2452 – 2459. doi: 10.1109/IROS.2011.6094759. Junique, C., Barbry, W., Scano, S., Zeliger, R., & Vergès, P. (2002). Ensembles de programmes permettant l’analyse de similitude de questionnaires et de données numeriques SIMI2000 (Manuel). [Bundle of programmes for the similitude analysis of questionaires and numerical data SIMI2000 (Manual)]. Kamide, H., Kawabe, K., Shigemi, S., & Arai, T. (2013). Development of a psychological scale for general impressions of humanoid. Advanced Robotics, 27 (1), 3-17. doi: 10.1080/01691864.2013.751159 Kleijnen, M., Lee, N., & Wetzels, M. (2009). An exploration of consumer resistance to innovation and its antecedents. Journal of Economic Psychology, 30 (3), 344-357. doi:10.1016/j.joep.2009.02.004 Lee, M., Forlizzi, J., Rybski, P., Crabbe, F., Chung, W., Finkle, J., & Kiesler, S. (2009, March). The snackbot: documenting the design of a robot for long-term humanrobot interaction. 4th ACM/IEEE International Conference on Human-Robot Interaction (HRI), 7-14. doi: 10.1145/1514095.1514100 Lee, S.-L., Lau, I., & Hong, Y.-y. (2011). Effects of Appearance and Functions on Likability and Perceived Occupational Suitability of Robots. Journal of Cognitive Engineering and Decision Making, 5 (2), 232 - 250. doi: 10.1177/1555343411409829. Lin, F., He, D., Jin, Y., Tao, Y., & Jiang, Z. (2013). Mapping the central structure core in social representation of pain. Journal of Applied Social Psychology, 43 (9), 1931–1945. doi: 10.1111/jasp.12135 Livingstone, S. (1992) The meaning of domestic technologies: a personal construct analysis of familial gender relations. In R. Silverstone and E. Hirsch (Eds.), Consuming Technologies. London: Routledge. ISBN 9780415069908Oudshoorn & Pinch 2003 Moliner, P. & Guimelli, C. (2015). Les représentation sociales. [Social representations]. Grenoble: Presses universitaires de Grenoble. Moliner, P. & Gutermann, M. (2004). Dynamique des descriptions et des explications dans une representation sociale. Papers on social representations - Textes sur les représentations sociales, 13, 2.1–2.12. Retrieved from http://www.psr.jku.at/PSR2004/13_02Mol.pdf Morritt, H. (1997). Women and Computer Based Technologies: A Feminist Perspective. Maryland, USA: University Press of America, Inc. Moscovici, S. (1961). La psychanalyse son image et son public [Psychoanalysis, its image and its public]. Paris, P.U.F. Mutlu, B. & Forlizzi, J. (2008). Robots in Organizations: The Role of Workflow, Social, and Environmental Factors in Human-Robot Interaction. HRI ’08, Proceedings of the 3rd ACM/IEEE international conference on Human robot interaction, 287-294. Retrieved from http://repository.cmu.edu/cgi/viewcontent.cgi? article=1036& context=hcii

Making sense of robot Nishio, S., Ishiguro, H., & Hagita, N. (2007). Geminoid: Teleoperated android of an existing person. In Armando C. P. Filho (Ed.), Humanoid Robots, New Developments (pp. 343-352). Vienna, Austria: I-Tech Education and Publishing. Retrieved form http://cdn.intechopen.com/pdfs-wm/240.pdf Oestreicher, L., & Eklundh, K. (2006). User Expectations on Human-Robot Cooperation. ROMAN ‘06, The 15th IEEE International Symposium on Robot and Human Interactive Communication, 91,96. doi: 10.1109/ROMAN.2006.314400 Oudshoorn, N. & T. Pinch (2003). How Users Matter: The Co-Construction of Users and Technologies. Cambridge/MA, London, the MIT Press. Piçarra, N., Giger, J.-C., Pochwatko, G., & Gonçalves, G. (2015). Validation of the Portuguese version of the Negative Attitudes towards Robots Scale. Revue Européenne de Psychologie Appliquée/European Review of Applied Psychology, 65, 93-104. Ram, S. & Sheth, J. (1989). Consumer resistance to innovations: the marketing problem and its solutions. Journal of Consumer Marketing, 6 (2), 5-14. doi: http://dx.doi.org/10.1108/EUM0000000002542 Ram, S. (1987). A model of innovation resistance. Advances in Consumer Research, 14 (1), 208 – 212. Retrieved from http://www.acrwebsite.org/search/view-conferenceproceedings.aspx?Id=6688 Ray, C., Mondada, F., & Siegwart, R. (2008). What do people expect from robots? IROS ‘08, International Conference on Intelligent Robots and Systems, 3816 3821. Retrieved from http://infoscience.epfl.ch/record/125291/files/iros08_ray_final-10.pdf Rindova, V. & Petkova, A. (2007). When Is a New Thing a Good Thing? Technological Change, Product Form Design, and Perceptions of Value for Product Innovations. Organization Science, 18 (2), 217 – 232. doi: 10.1287/orsc.1060.0233 Rogers (1983). Diffusion of innovations, 3rd edition. New York: Free Press. Rouquette, M.-L. & Rateau, P. (1998). Introduction à l'étude des représentations sociales [An introduction to the study of social representations]. Grenoble: Presses universitaires de Grenoble. . Sabanovic, S. (2010). Robots in Society, Society in Robots. Mutual Shaping of Society and Technology as a Framework for Social Robot Design. International Journal of Social Robotics, 2, (4), 439 - 450. doi: 10.1007/s12369-010-0066-7 Salès-Wuillemin, E., Morlot, R., Fontaine, A., Pullin, W., Galand, C., Talon, D., & Minary-Dohen, P. (2011). Evolution of nurses’ social representations of hospital hygiene: From training to practice. Revue européenne de psychologie appliquée, 61 (1), 51-63. doi:10.1016/j.erap.2010.06.001 Scopelliti, M., Giuliani, M., & Fornara, F. (2005). Robots in a domestic setting: a psychological approach. Universal Access in the Information Society, 4 (2), 146155. doi: 10.1007/s10209-005-0118-1

Making sense of robot Shenkar, O. (1988). Robotics: A challenge for occupational psychology. Journal of Occupational Psychology, 61, 103-112. Sheth, J. (1981). Psychology of innovation resistance: the less developed concept (LDC) in diffusion research. Research in Marketing, 4 (3), 273-282. Retrieved from http://www.bus.emory.edu/jSheth/docs/Psychology%20of%20Innovation%20Resis tance.pdf Takayama, L., Ju, W., & Nass, C. (2008). Beyond Dirty, Dangerous and Dull: What Everyday People Think Robots Should Do. Proceedings of the 3rd ACM/IEEE international conference on Human robot interaction, 25 – 32. doi: 10.1145/1349822.1349827 TNS Opinion & Social (2012). Public attitudes towards robots (Special Eurobarometer 382). European Comission. Retrieved from http://ec.europa.eu/public _opinion/index_en.htm Turkle, S. (1988). Computational Reticence: Why Women Fear the Intimate Machine, in C. Kramarae (ed.), Technology and Women’s Voices. London: Routledge and Kegan Paul. Turkle, S., Breazeal, C., Dasté, O., & Scassellati, B. (2006). Encounters with kismet and cog: Children respond to relational artifacts. Digital Media: Transformations in Human Communication, 1-20. Retrived from http://scazlab.yale.edu/publications/ all-publications Vala, J. (1993). Representações sociais – para uma psicologia social do pensamento social.[Social representations - for a psychology of social thought] In J. Vala e M. B. Monteiro (Ed.) Psicologia Social (pp. 353-384). Lisboa: Fundação Calouste Gulbenkian. Veltri, N. F., Krasnova, H., Baumann, A., Kalayamthanam, N. (2014). Gender Differences in Online Gaming: A Literature Review, Conferences papers, Twentieth Americas Conference on Information Systems, Savannah, USA. Vergès, P. (1992). L’évocation de l’argent. Une méthode pour la définition du noyau central d’une représentation [The evocation of money. A method for the identification of the central nucleus of a representation]. Bulletin de Psychologie, 405 (45), 203–209. Vergès, P. (2001). L’analyse dês représentations sociales par questionnaires. [The analysis of social representations using questionnaires]. Revue Française de Sociology, 42(3), 537-561. http://dx.doi.org/10.2307/3323032 Vergès, P., Scano, S., & Junique, C. (2002). Ensembles de programmes permettant l’analyse des évocations EVOC2000 (Manuel). [Bundle of programs for the analysis of evocations EVOC2000 (Manual)] Aix en Provence: Université Aix en Provence. Vergès, P., Tyszka, T., & Vergès, P. (1994). Noyau central, saillance et propriétés structurales. Papers on social representations – textes sur les représentations sociales, 3, 3-12. Retrieved from http://www.psych.lse.ac.uk/psr/

Making sense of robot Young, J. E., Hawkins, R., & Sharlin, E. (2009). Toward Acceptable Domestic Robots: Applying Insights from Social Psychology. International Journal of Social Robotics, 1 (1), 95-108. doi: 10.1007/s12369-008-0006-y Young, J., Sung, J., Voida, A., Sharlin, E., Igarashi, T., Christensen, H., & Grinter, R. (2011). Evaluating Human-Robot Interaction. Focusing on the Holistic Interaction Experience. International Jounal of Social Robotics, 3, 53 –67. doi: 10.1007/s12369-010-0081-8

Making sense of robot

Table 1 Socio-demographic characteristics of the participants in study 1 Study 1 Age M SD Min-Max Not reported Gender Female Male Not reported Years of School Up to 9 Up to 12 University Degree Not reported

N=212 34.01 12.58 19-69 16 128 76 8 67 23 116 6

Occupation Student Management, sales & public service Education & Health Engineering Construction Tourism Unemployed Other Not reported

108 26 9 8 26 26 5 4

Making sense of robot

Table 2 Evocation frequency distribution Frequency Nº of words Cumulative evocations 1 413 413 24.8 % 2 56 525 31.5 % 3 29 612 36.7 % 4 15 672 40.3 % 5 14 742 44.5 % 6 3 760 45.6 % 7 1 767 46.0 % 8 4 799 48.0 % 9 5 844 50.7 % 10 4 884 53.1 % 11 3 917 55.0 % 12 3 953 57.2 % 13 2 979 58.8 % 14 1 993 59.6 % 16 2 1025 61.5 % 17 3 1076 64.6 % 18 1 1094 65.7 % 19 1 1113 66.8 % 20 1 1133 68.0 % 21 2 1175 70.5 % 22 1 1197 71.8 % 24 1 1221 73.3 % 27 1 1248 74.9 % 30 1 1278 76.7 % 31 1 1309 78.6 % 32 2 1373 82.4 % 36 1 1409 84.6 % 38 1 1447 86.9 % 45 1 1492 89.6 % 74 1 1566 94.0 % 100 1 1666 100.0 % Notes: Minimum frequency and mean frequency in bold

Cumulative inversed 1666 100.0 % 1253 75.2 % 1141 68.5 % 1054 63.3 % 994 59.7 % 924 55.5 % 906 54.4 % 899 54.0 % 867 52.0 % 822 49.3 % 782 46.9 % 749 45.0 % 713 42.8 % 687 41.2 % 673 40.4 % 641 38.5 % 590 35.4 % 572 34.3 % 553 33.2 % 533 32.0 % 491 29.5 % 469 28.2 % 445 26.7 % 418 25.1 % 388 23.3 % 357 21.4 % 293 17.6 % 257 15.4 % 219 13.1 % 174 10.4 % 100 6.0 %

Making sense of robot Table 3 The four quadrant diagram representing frequency and order of evocation Frequency ≥13 Mean Order