Evaluating Mentalization during Driving

Giorgio Grasso, Chiara Lucifora, Pietro Perconti and Alessio Plebe

Department of Cognitive Science, University of Messina, Italy

Keywords:

Autonomous Driving, Mindreading, Eye Contact.

Abstract:

The development of artiﬁcial intelligence promises important future changes from a social point of view. In

particular, the emerging self-driving cars allow today to plan a future where trafﬁc ﬂow will greatly improve,

and car accidents will be continuously decreasing. However, we should expect a period when full or partial

autonomous vehicles and ordinary cars coexist, during which it would be essential to fully understand the

cognitive processes used by ordinary people when driving. We identify as a crucial aspect the shift between

quick and automated reactions, and the resort to mentalizing, costly social processes, sometimes necessary

to predict intentions of other road users. In our experimental design we investigate the main precursors of

mindreading, that is, eye contact and shared attention. We believe that a better understanding of this twofold

mecahnisms involved in driving could be used to improve advanced driver assistance systems.

1 INTRODUCTION

Since few years Artiﬁcial Intelligence is enjoying its

most fortunate period ever (Schwab, 2016; Parloff,

2016; Makridakis, 2017). The astonishing advances

achieved since, are changing the way many problems

are solved to the extent that the world is said to experi-

ence a AI Renaissance (Tan and Lim, 2018). This dra-

matic progress is almost entirely due to artiﬁcial neu-

ral networks in their new deep versions (Schmidhu-

ber, 2015; Goodfellow et al., 2016; Chui et al., 2018;

Hazelwood et al., 2018).

Research on autonomous vehicles is certainly one

of the application areas that mostly beneﬁted from

the rise of the deep learning (Gurghian et al., 2016;

Wu et al., 2016; Rausch et al., 2017; Bojarski et al.,

2017; Li et al., 2018; Schwarting et al., 2018). Adop-

tion of deep neural models in the automotive domain

is now attainable by exploiting graphics processing

units (GPUs), thanks to the CUDA software interface

(Sanders and Kandrot, 2014), and real-time GPU-

based computers like NVIDIA Drive PX. Together

with technologies enabling dedicated communication

(V2x) between vehicles and other vehicles and infras-

tructures (Zhao et al., 2018), a future of monopoly of

algorithms over trafﬁc will be real, with great advan-

tages in terms of trafﬁc efﬁciency and safety.

Notwithstanding, the shift from human to auto-

matic driving will pose serious challenges, for a se-

riers of crucial issues, like readiness of infrastruc-

tures (Johnson, 2017). Therefore, in the most prob-

able near future full or partial autonomous vehicles

must coexist with ordinary non-autonomous vehicles.

Bearing this in mind, an important question concerns

the cognitive processes used by ordinary people when

driving, in particular during critical interactions with

other active agents, such as cars, trucks, cyclists,

and pedestrians. Several researches, reviewed in the

next Section, are addressing this question, which still

remains largely unanswered. Our work attempt to

progress in this direction, by designing a system for

assessing when and why subjects resort to costly so-

cial processes, rather than using quick and automated

reactions.

By drawing on current cognitive science and neu-

roscience, is it possible to identify number of differ-

ent brain processes recruited when driving, from low-

level sensorimotor coordination up to social cognition

and decision making. In particular, it will be argued

for the idea that we have to take into account mental-

izing abilities (Samson, 2013; Vilarroya and i Argi-

mon, 2007) i.e., social cognition processes aiming at

inferring intentions of others and not simply at paying

attention to other people behavior by means of au-

tomated sensorimotor control processes. As we will

brieﬂy review in §2.1, cognitive science is not an ob-

vious perspective in the assesment of driver’s behav-

ior, there are several other approaches, some of which

just neglect the cognitive aspect, few others explicity

foster an alternative perspective.

Our experimental design is intended to test the

typical perceptual conditions which elicit the shift

536

Grasso, G., Lucifora, C., Perconti, P. and Plebe, A.

Evaluating Mentalization during Driving.

DOI: 10.5220/0007756505360541

In Proceedings of the 5th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2019), pages 536-541

ISBN: 978-989-758-374-2

from an automated driving style to a mentalizing as-

sessment of the driving scene. We test the main pre-

cursors of mindreading, that is, eye contact and shared

attention (Tomasello, 2009a; Shepherd, 2010). We ar-

gued for the thesis that measurable eye contact and

joint attention perceptual patterns are typical of sce-

narios which elicit social cognition and mentalizing

drive. Our attempt ends up in suggesting that these

empirical ﬁndings could be used to improve advanced

driver assistance systems in the cases in which a men-

talizing driving style is needed.

2 MENTALIZATION AND

DRIVING

Driving is a well disciplined human activity, with

rules covering almost every possible pattern of in-

teractions between vehicles, and between other non

driving people. Driving rules are designed to opti-

mize trafﬁc ﬂow and maximize road safety. There is

a continuous trend of harmonization of driving rules

among countries (WP.29, 2012), even if the situation

is highly unbalanced between high- and mid- income,

and low-income countries (WHO, 2018). Therefore,

for a driver with enough experience it might sufﬁce to

follow the rules to interact efﬁciently with the other

human actors on the road, and this rule-following be-

havior would usually require automated sensorimotor

controls only. However, blind trusting the rules may

turn out highly dangerous, and often drivers should

resort to a forecast of the next actions of others, in-

cluding possible actions that violate road rules.

This sort of activity belong to what is generally

called “mindreading”, one of the most enhanced hu-

man cognitive capacity, ﬁrst noticed by social psy-

chologists and philosophers in the middle of last cen-

tury (Sellars, 1956; Heider, 1958), since then grown

as a central topic in philosophy of mind (Dennett,

1987; Nichols and Stich, 2003; Goldman, 2006;

Hutto, 2008; Perconti, 2017) and cognitive science

(Stich, 1983; Gopnik and Meltzoff, 1997; Gopnik and

Meltzoff, 1999; Brooks and Meltzoff, 2002).

2.1 Mindreading and Its Biological

Basis

Generally speaking, mindreading is a set of cognitive

capacities that allow people to predict others’ behav-

ior in a wide range of circumstances; that encour-

age to attribute mental states to humans; and to ex-

plain the behavior of humans in terms of their pos-

sessing mental states. There are several open dis-

cussions, for example if mindreading is based on a

sort of simpliﬁed theory of human behavior repre-

sented in our brains (Gopnik and Meltzoff, 1997;

Gopnik and Meltzoff, 1999; Gopnik and Wellman,

2012; Samson, 2013), or if it can better be conceived

as a sort of simulation (Gallese and Goldman, 1998;

Goldman, 2006), the same kind of strategy broadly

adopted in high level cognition (Hesslow, 2012). De-

spite open problems, there is now an overall com-

prehensive and detailed picture of how mindreading

works, its evolutionary reasons (Call and Tomasello,

2005; Tomasello, 2009b), and its neural correlates

(Rizzolatti et al., 2001; Umilt

a et al., 2001; Spiers and

Maguire, 2006). One clear aspect is that mindreading

is cognitively expensive, therefore is used with parsi-

mony. Generally speaking, human behavior can be

characterized by low or high levels of mentalizing.

Above all, it depends on how much they are automatic

or voluntary. It is, however, also matter of how much

the behavior is social. Swimming alone in a swim-

ming pool is a typical low level mentalizing behavior.

It is, in fact, a non-social and automatic action. On

the contrary, when someone is asked to intercede to

pacify two contenders, it is necessarily engaged in a

high level mentalizing behavior. It is an intentional

action and it is directed to another individual and to

a social scenario. It is not only matter of attention

and will. It depends, in fact, on interpreting (or not)

a given behavior by means of the intentional vocabu-

lary and the folk psychology framework, consisting of

believes, desires, and propositional attitudes. Mental-

izing can be more or less useful when you are driving

a vehicle. Driving a car in an isolated motorway is

a completely different experience than trying, to say,

to cross a road in downtown Hanoi. Guessing other

people intentions is a crucial issue in one case, but

not in the other. For this, mental processes which are

involved during driving are so various. Sometimes so-

cial cognition processing is highly demanding, while

in other circumstances the brain works, so to speak,

in a solipsistic and automatic way. Trafﬁc rules play

a key role in leading people to adopt a more or less

mentalizing driving style. What matters is the conve-

nience to adopt a negotiation driving style, or to base

our behavior merely on conventional rules and habits.

When we will be able to fully model how that “con-

venience” works, we will be endowed with a fruitful

theoretical resource to better deal with the social au-

tonomous vehicle challenges.

Trying to discriminate the neural basis of spon-

taneous vs. voluntary mentalizing during everyday

experiences, (Spiers and Maguire, 2006) used as a

case study for their investigation the taxi driver or-

dinary experience of driving in central London. They

Evaluating Mentalization during Driving

537

found an increased activity in a number of regions,

namely the right pSTS, the mPFC and the right tem-

poral pole, largely overlapping with many neuroimag-

ing studies examining the neural basis of mentalizing.

It seems that when the driver shifts from the “coast-

ing driving style, where subjects were actively driv-

ing and moving through the city, but did not have any

directed thoughts”, to a negotiating driving style, the

brain starts to work in a highly mentalizing mode.

2.2 Relevant Researches

Despite the current improved understanding of how

mentalizing works, as just reviewed, ans the impor-

tance during driving, there are few relevant studies so

far. There is an increasing number of studies on so-

cial interaction between road users, for the purpose

to implement communicative devices in autonomous

vehicles. (Riaz and Niazi, 2017) present a social au-

tonomous vehicle (AV), with the capability of predict-

ing intentions, mentalizing and copying the actions of

each other. Their cognitive architecture includes two

modules: The Mentalizing Module in charge of dis-

covering the intention of neighboring vehicles, and

the Mirroring Module, in charge of changing the ego

trajectory according to the changed trajectory of the

nearest vehicles. Although Riaz and Niazi maintain

that their model will improve the collision avoidance

capabilities of AVs, it does not try to simulate the cog-

nitive faculties actually involved in performing the

given task. Anthropomorphism, a process whereby

people attribute to nonhumans distinctively human

characteristics, particularly the capacity for planing

and taking decisions, is the focus of the study by

(Waytz et al., 2014), still aimed at identify poten-

tial advantages for autonomous vehicle. The work of

(Zhang et al., 2018) is also aimed at proposing new

forms of vehicle communication signals, with the pur-

pose of indicating its “intentions”.

Several researches have addressed the question

of how drivers predict the behavior of pedestrians

(Jorge and Rossetti, 2018; Bengtsson, 2018; Rasouli

et al., 2018), this limited target is justiﬁed because for

pedestrians only the common strategies of mentaliz-

ing like eye gazing can be applied. However, drivers

do attempt to apply their normal mentalizing strate-

gies to other cars too, even if the scarce visibility of

the head of other drivers reduce drastically the efﬁ-

ciency of mentalizing. Moreover, even when limited

to pedestrians, as far as our knowledge all researches

have failed to distinguish between the application of

the most common automated sensorimotor control,

and the switch to the more costly mentalizing.

While the body of studies reviewed so far, even

if not aimed at evaluating mentalizing per se in the

road context, do assume it is the main behavior at

play, there are several other studies that deliberately

disregard the perspective of mentalizing. This choice

is grounded in the same old antipsychologism typi-

cal of the analytic philosophy of last century, as clari-

ﬁed, for example, by (Broth et al., 2018): “Since Ryle

(1949) and Wittgenstein (1953), many action-oriented

scholars have discussed how ‘understanding’ another

person rests not on having (mediated or direct) ac-

cess to their mind – which is presumably lodged in

the brain – but on what they are relevantly doing in

a particular situation. [. . . ] Respeciﬁed as a mani-

fest and social phenomenon rather than as an internal

and private one, intentions can be studied as they are

oriented to by participants over the course of inter-

actional sequences”. Adhere to this viewpoint sev-

eral studies focused on analyzing interactions of road

users using the framework of ethnomethodology and

conversation analysis (Merlino and Mondada, 2018).

In the study by (Broth et al., 2018), quoted above,

ascriptions of mental states to other road users are ex-

amined in the context of driver training. A similar

non-cognitive framework is that of micro sociology,

founded on the approach of Erving (Goffman, 1963),

inspired the recent approach called “mobile ethnog-

raphy” for the study of interactions of people related

to mobility negotiation (Jensen, 2010; Karndacharuk

et al., 2014). Studies within this framework encom-

passes speciﬁc aspects of social interactions between

road users such as ways to communicate the inten-

tion to offer space to each other (Haddington and Rau-

niomaa, 2014), or the analysis of overt appreciations

of the actions of other drivers (Laurier, 2018).

The human factors domain is an additional rel-

evant research ﬁeld, with a long tradition in the

study of driver behavior, where often psychologi-

cal categories are used, such as attention (Kircher

and Ahlstrom, 2017) or engagement (Radwin et al.,

2017), therefore in between a behavioral approach

and our persepective.

2.3 Our Approach

Unlike the studies reviewed above, our approach is

speciﬁcally aimed at discriminating when subjects

make use of mentalizing, in a simulated driving sce-

nario. Therefore, not only we do assume e cogni-

tive viewpoint, but it is precisely the discrimination

between automatic/mentalizing states the objective of

our study. Our experimental design is intended to test

the typical perceptual conditions which elicit the shift

from an automated driving style to a mentalizing as-

sessment of the driving scene. We test the main pre-

VEHITS 2019 - 5th International Conference on Vehicle Technology and Intelligent Transport Systems

538

Figure 1: Example of trafﬁc situations that may engage the subjexct in mentalizing.

cursors of mindreading, that is, eye contact and shared

attention. We argued for the thesis that measurable

eye contact and joint attention perceptual patterns are

typical of scenarios which elicit social cognition and

mentalizing drive.

There is ample scientiﬁc support that suggests eye

monitoring as one of the most reliable source of in-

formation about how a subject is actually engaged in

mentalizing (Wiese et al., 2012; Turner and Felisberti,

2017). It appears that distinct neural systems have

evolved to process two crucial types of gaze informa-

tion: direct and deictic gaze (Shepherd, 2010). The

former is associated with the likelihood that an indi-

vidual will engage the observer. Deictic gaze signals

spatial attention, suggests future actions, and deﬁne

potential shared targets.

Our attempt ends up in suggesting that these em-

pirical ﬁndings could be used to improve advanced

driver assistance systems in the cases in which a men-

talizing driving style is needed.

3 THE DRIVING SIMULATOR

Our experimental design was developed with the aim

of recreating a realistic environment to simulate driv-

ing in an urban context, populated by cars, pedestrians

and other typical object and visual clues characteriz-

ing a urban scenery. The driving simulator was devel-

oped using a software environment largely employed

for vieogames (Unity) and an hardware setup, com-

posed of immersive virtual reality monitors, a Log-

itech steering wheel with force feedback, pedals and

gearshift used to provide a complete driving experi-

ence during the simulation. The main measure is the

eye gazing of the subjects when driving, using To-

bii eye tracking integrated within Unity, in order to

project eye gazing into the driving scene.

A highly detailed city model has been employed,

taken from the free repository of Unity, namely the

Windridge City for AirSim on Unity. The package

includes urban roads surrounded by forest and extra-

urban roads; interconnected roads; outdoor furniture,

trafﬁc signs and buildings. The model comprises a

proper city environment and includes an extra-urban

road, which allows the subject to get acquainted with

the simulation in the absence of trafﬁc.

We have extensively worked on pedestrians and

vehicles, in order to adapt available models to the

needs of the simulation. In particular pedestrians have

been animated to simulate a typical walkabout behav-

ior. The simulated vehicle is controlled by a dynamic

model, which includes the physical characteristics of

real vehicles, in terms of inertia, friction, collision de-

tection and engine traction. The vehicle returns to the

user shock feedback, via the steering movement. The

steering modiﬁes its resistance in relation to car speed

and asphalt conditions.

Pedestrians of male and female gender are

equipped with a RigidBody component that allows

them to obey the laws of physics during movem,ent,

to receive shocks, to be subject to gravity and to sim-

ulate friction and collisions with other objects. If the

pedestrians are hit by vehicle, they may fall down,

as in reality. Through NavMesh, it has been possi-

ble to divide the city into pedestrian areas (sidewalks)

and non-pedestrian areas (carriageways). Inside the

pedestrian areas the pedestrians move through appro-

priate Waypoints, their animation is regulated by the

the iskinematic option, available in the Unity Simu-

lation Engine. The vehicle route is also controlled

via the NavMesh package and waypoints. Each vehi-

cle has a Collider component, speciﬁcally the Mesh-

Collider has been used to compute collisions, based

on an effective object’s geometry. Vehicles can be

divided into two categories. One that can circulate

freely within the city, while the other follow a pre-

established path, that allows them to cross the driver’s

path. Only the latter has a controlled speed, whose

module is multiplied by an appropriate scale to match

the user’s speed. Some vehicles are also connected to

an invisible trigger, which allows the activation of a

speciﬁc behavior of pedestrians and/or other vehicles

at the passage of the user’s vehicle.

Examples of the view from inside the virtual car

are shown in Fig. 1, in situations where the subject

Evaluating Mentalization during Driving

539

Figure 2: Example of high mindreading event.

can either use automatic control, trusting trafﬁc rules,

or shift to inferring the intentions of other road users,

independently from rule following, by mentalizing.

There are situations designed to forse mentalization,

as in Fig. 2, when suddenly a child crosses the street

without looking at the incoming vehicle.

REFERENCES

Bengtsson, P. (2018). Attuning the pedestrian-vehicle and

driver-vehicle – why attributing a mind to a vehicle

matters. In Karwowski, W. and Ahram, T., editors, In-

ternational Conference on Intelligent Human Systems

Integration, pages 308–319.

Bojarski, M., Yeres, P., Choromanaska, A., Choromanski,

K., Firner, B., Jackel, L., and Muller, U. (2017). Ex-

plaining how a deep neural network trained with end-

to-end learning steers a car. CoRR, abs/1704.07911.

Brooks, R. and Meltzoff, A. N. (2002). The importance of

eyes: how infants interpret adult looking behaviour.

Developmental Psychology, 38:958–966.

Broth, M., Cromdal, J., and Levin, L. (2018). Telling the

other’s side. formulating others’ mental states in driver

training. Language & Communication.

Call, J. and Tomasello, M. (2005). What chimpanzee

know about seeing, revisited: An explanation of the

third kind. In Eilian, N., Hoerl, C., McCormack, T.,

and Roessler, J., editors, Joint Attention: communica-

tion and other minds, pages 45–64. Oxford University

Press, Oxford (UK).

Chui, M., Manyika, J., Miremadi, M., Henke, N., Chung,

R., Nel, P., and Malhotra, S. (2018). Notes from the AI

frontier: insights from hundreds of use cases. Techni-

cal Report April, McKinsey Global Institute.

Dennett, D. C. (1987). The Intentional Stance. MIT Press,

Cambridge (MA).

Gallese, V. and Goldman, A. (1998). Mirror neurons and the

simulation theory of mind-reading. Trends in Cogni-

tive Sciences, 2:493–501.

Goffman, E. (1963). Behavior in public places: Notes on

the social organisation of gatherings. Free Press, New

York.

Goldman, A. (2006). Simulating Minds: The Philosophy,

Psychology, and Neuroscience of Mindreading. Ox-

ford University Press, Oxford (UK).

Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep

Learning. MIT Press, Cambridge (MA).

Gopnik, A. and Meltzoff, A. N. (1997). Words, Thoughts

and Theories. Oxford University Press, Oxford (UK).

Gopnik, A. and Meltzoff, A. N. (1999). The Scientist in the

Crib. HarperCollins, London.

Gopnik, A. and Wellman, H. M. (2012). Reconstruct-

ing constructivism: Causal models, bayesian learn-

ing mechanisms, and the theory-theory. Psychological

Bulletin, 138:1085–1108.

Gurghian, A., Koduri, T., Bailur, S. V., Carey, K. J., and

Murali, V. N. (2016). DeepLanes: End-to-end lane

position estimation using deep neural networks. In

Proc. of IEEE International Conference on Computer

Vision and Pattern Recognition, pages 38–45.

Haddington, P. and Rauniomaa, M. (2014). Interaction be-

tween road users: Offering space in trafﬁc. Space and

Culture, 17:176–190.

Hazelwood, K., Bird, S., Brooks, D., Chintala, S., Diril, U.,

Dzhulgakov, D., Fawzy, M., Jia, B., Jia, Y., Kalro, A.,

Law, J., Lee, K., Lu, J., Noordhuis, P., Smelyanskiy,

M., Xiong, L., and Wang, X. (2018). Applied machine

learning at Facebook: A datacenter infrastructure per-

spective. In IEEE International Symposium on High

Performance Computer Architecture (HPCA), pages

620–629.

Heider, F. (1958). The Psychology of Interpersonal Rela-

tions. John Wiley, New York.

Hesslow, G. (2012). The current status of the simulation

theory of cognition. Brain, 1428:71–79.

Hutto, D. D. (2008). Folk Psychological Narratives: The

Sociocultural Basis of Understanding Reasons. MIT

Press, Cambridge (MA).

Jensen, O. B. (2010). Negotiation in motion: Unpacking a

geography of mobility. Space and Culture, 13:389–-

402.

Johnson, C. (2017). Readiness of the road network for con-

nected and autonomous vehicles.

Jorge, C. C. and Rossetti, R. J. F. (2018). On social interac-

tions and the emergence of autonomous vehicles. In

International Conference on Vehicle Technology and

Intelligent Transport Systems, pages 423–430.

Karndacharuk, A., Wilson, D. J., and Dunn, R. (2014). A re-

view of the evolution of shared (street) space concepts

in urban environments. Transport reviews, 34:190–

220.

Kircher, K. and Ahlstrom, C. (2017). Minimum required

attention: A human-centered approach to driver inat-

tention. Human factors, 59:471–484.

Laurier, E. (2018). Civility and mobility: Drivers (and pas-

sengers) appreciating the actions of other drivers. Lan-

guage & Communication.

Li, J., Cheng, H., Guo, H., and Qiu, S. (2018). Survey on

artiﬁcial intelligence for vehicles. Automotive Innova-

tion, 1:2–14.

VEHITS 2019 - 5th International Conference on Vehicle Technology and Intelligent Transport Systems

540

Makridakis, S. (2017). The forthcoming artiﬁcial intelli-

gence (AI) revolution: Its impact on society and ﬁrms.

Futures, 90:46–60.

Merlino, S. and Mondada, L. (2018). Crossing the street:

How pedestrians interact with cars. Language & Com-

munication.

Nichols, S. and Stich, S. P. (2003). Mindreading: An Inte-

grated Account of Pretence, Self-awareness, and Un-

derstanding Other Minds. Oxford University Press,

Oxford (UK).

Parloff, R. (2016). Why deep learning is suddenly changing

your life. Fortune, September:28.

Perconti, P. (2017). Filosoﬁa della mente. Il Mulino,

Bologna (IT).

Radwin, R. G., Lee, J. D., and Akkas, O. (2017). Driver

movement patterns indicate distraction and engage-

ment. Human factors, 59:844–860.

Rasouli, A., Kotseruba, I., and Tsotsos, J. K. (2018). Un-

derstanding pedestrian behavior in complex trafﬁc

scenes. IEEE Transactions on Intelligent Vehicles,

3:61–70.

Rausch, V., Hansen, A., Solowjow, E., Liu, C., Kreuzer,

E., and Hedrick, J. K. (2017). Learning a deep neural

net policy for end-to-end control of autonomous vehi-

cles. In Proceedings of American Control Conference,

pages 4914–4919.

Riaz, F. and Niazi, M. (2017). Towards social autonomous

vehicles: Efﬁcient collision avoidance scheme us-

ing Richardson’s arms race model. PLoS ONE,

12:e0186103.

Rizzolatti, G., Fogassi, L., and Gallese, V. (2001). Neuro-

physiological mechanisms underlying the understand-

ing and imitation of action. Nature Reviews Neuro-

science, 2:661–670.

Samson, D. (2013). Theory of mind. In Reisberg, D., ed-

itor, The Oxford Handbook of Cognitive Psychology.

Oxford University Press, Oxford (UK).

Sanders, J. and Kandrot, E. (2014). CUDA by example: an

introduction to general-purpose GPU programming.

Addison Wesley, Reading (MA).

Schmidhuber, J. (2015). Deep learning in neural networks:

An overview. Neural Networks, 61:85–117.

Schwab, K. (2016). Fourth Industrial Revolution. World

Economic Forum, Geneva.

Schwarting, W., Alonso-Mora, J., and Rus, D. (2018). Plan-

ning and decision-making for autonomous vehicles.

Annual Review of Control, Robotics, and Autonomous

Systems, 1:8.1–8.24.

Sellars, W. (1956). Empiricism and the philosophy of

mind. In Feigl, H. and Scriven, M., editors, Minnesota

Studies in the Philosophy of Science, volume I: The

Foundations of Science and the Concepts of Psychol-

ogy and Psychoanalysis, pages 253–329. University

of Minnesota Press.

Shepherd, S. V. (2010). Following gaze: gaze-following

behavior as a window into social cognition. Frontiers

in Integrative Neuroscience, 4:5.

Spiers, H. and Maguire, E. A. (2006). Spontaneous men-

talizing during an interactive real world task: an fMRI

study. Neuropsychologia, 44:1674–1682.

Stich, S. P. (1983). From Folk Psychology to Cognitive Sci-

ence. MIT Press, Cambridge (MA).

Tan, K.-H. and Lim, B. P. (2018). The artiﬁcial intelligence

renaissance: deep learning and the road to human-

level machine intelligence. APSIPA Transactions on

Signal and Information Processing, 7:e6.

Tomasello, M. (2009a). Origins of human communication.

MIT Press, Cambridge (MA).

Tomasello, M. (2009b). Why We Cooperate. MIT Press,

Cambridge (MA).

Turner, R. and Felisberti, F. M. (2017). Measuring min-

dreading: A review of behavioral approaches to test-

ing cognitive and affective mental state attribution in

neurologically typical adults. Frontiers in Psychology,

8:47.

Umilt

a, M., Kohler, E., Gallese, V., Fogassi, L., Fadiga, L.,

Keysers, C., and Rizzolatti, G. (2001). I know what

you are doing – a neurophysiological study. Neuron,

31:155–165.

Vilarroya, O. and i Argimon, F. F., editors (2007). Social

Brain Matters – Stances on the Neurobiology of Social

Cognition. Rodopi, Amsterdam.

Waytz, A., Heafner, J., and Epley, N. (2014). The mind in

the machine: Anthropomorphism increases trust in an

autonomous vehicle. Journal of Experimental Social

Psychology, 52:113–117.

WHO (2018). Global status report on road safety.

Wiese, E., Wykowska, A., Zwickel, J., and M

uller, H. J.

(2012). I see what you mean: How attentional selec-

tion is shaped by ascribing intentions to others. PLoS

ONE, 7:e45391.

WP.29 (2012). World forum for harmonization of vehicle

regulations.

Wu, B., Iandola, F. N., Jin, P. H., and Keutzer, K. (2016).

SqueezeDet: Uniﬁed, small, low power fully convo-

lutional neural networks for real-time object detection

for autonomous driving. CoRR, abs/1612.01051.

Zhang, J., Vinkhuyzen, E., and Cefkin, M. (2018). Evalu-

ation of an autonomous vehicle external communica-

tion system concept: a survey study. In Stanton, N. A.,

editor, Advances in Human Aspects of Transportation,

pages 650–661. Springer-Verlag, Berlin.

Zhao, Y., Yao, S., Shao, H., and Abdelzaher, T. (2018). Co-

drive: Cooperative driving scheme for vehicles in ur-

ban signalized intersections. In ACM/IEEE Interna-

tional Conference on Cyber-Physical Systems, pages

308–319.

Evaluating Mentalization during Driving

541