Introduction

Empathy has been defined as the state where perceivers represent the same emotion they are observing or imagining in another person with full awareness that the source of their own experience is the other’s emotion.¹ All factors that constitute, or stem from, the concept of empathy have been explored under various conditions in neuroscience. Such enthusiastic exploration of the mechanisms of empathy reflects the perception that empathy is generally understood to be the basis of humans’ prosocial nature and the cognitive capacity needed for altruistic behavior.^2–4 Taking into account the requirement for investigation of empathy based on social interests and real-life issues, the present study focused on relationships between empathy and depictions in a fiction film, an issue which has rarely been investigated. By measuring brain activity of participants viewing a fiction film under natural viewing conditions with no experimental interventions, the present study examines and discusses the interactive process involved in generation of empathy for the film.

Using a methodologically inventive approach shared by Hasson et al⁵ and Bartels and Zeki,⁶ natural viewing conditions were adopted for this experiment in order to avoid the potential problem of using data obtained from an experiment performed under limiting conditions in order to discuss cognitive states present during natural viewing. Thus, the present study was conducted based on the hypothesis that the mirror neuron system (MNS), which is activated by a network of neurons in multiple brain areas involved in the motor function of the subject when the subject performs an action and observes the same action being performed by the target, is the main neural basis for emotional empathy.^7–9 Overlap of cortical activation during self-produced actions and actions observed from others can be interpreted as “isomorphic” affective states; that is, sharing of the same emotion by the perceiver and others.^10–14 Research in MNS suggest that actions and facial expressions are generally considered to be the main information to be provided by others to evoke such isomorphic affective states in the perceiver.^15,16 However, based on a presumption that the MNS activity can be induced during the viewing of a fiction film by information obtained not only from relevant actions and facial expressions of characters in the film but also from elements that do not belong to the characters, such as story and directorial technique, the present study intentionally used, as the material of the experiment, a nonexplanatory fiction film in which characters show only minimal actions or facial expressions while providing intentionally limited lingual information that might complement their constrained actions or facial expressions.

“Minimization of explanatory elements” is a common world-class filmmaking technique, which is supported by internationally respected film directors. How do mirroring related to the MNS and neural activation that infers the mental state of others (mentalizing) interact to construct cognition in the brain of the perceiver viewing a film with minimal explanatory elements? To investigate this point, we employed Dolls, directed by Takeshi Kitano (http://www.office-kitano.co.jp/dolls/), which was judged to be the most suitable material for the present study because of its tragic love story which would likely induce empathy for the characters while explanatory elements were used as little as possible.

In associated fields such as psychology and cognitive science, only a few studies of empathy in fiction films have been conducted.^17–19 Meanwhile, in the field of neuroscience, a small core of researchers has been researching this issue intensively, building a solid foundation for further research.^5,20–22 Most natural viewing studies emphasize the necessity of doing experiments without strict constraints, in which participants are allowed to focus naturally on the presented stimulus, while most of those utilizes functional magnetic resonance imaging. In the present study, we use near-infrared spectroscopy (NIRS), which is a noninvasive neuroimaging method that measures changes in the oxyhemoglobin (oxy-Hb), deoxyhemoglobin (deoxy-Hb), and total-hemoglobin (total-Hb) concentrations.²³ NIRS has been applied to various cognitive studies, including sensorimotor, visual perception, language, and executive functions; developmental studies; and clinical studies.^24–28 NIRS makes it possible to measure brain activity in a nearly “natural state” with almost no physical constraints given to participants. This is the methodological superiority that is not fulfilled by the experiment that utilized functional magnetic resonance imaging.

Our greatest concern here is not to generally specify the activated brain regions during film viewing but to show the process of empathy by the data extracting overlap of fictional characters and perceiver’s motor cortex. The standpoint of the present study is based on the knowledge that MNS activity contributes to represent a person’s action that in turn leads to a deeper understanding of the internal states such as intentions or emotions of another person.²⁹ Moreover, the present study explores how neuroscience can be used to investigate films further in an artistically valid manner. The methodology of neuroscience is expected to provide useful suggestions as to how films are studied.

Methods

Participants

Twelve healthy adult individuals (six females and six males, aged 19–23 years; mean ± standard deviation: 21.08±1.44 years) participated in the experiment. None of these participants had watched this experimental material before, and all but one did not know the film, which was confirmed on the basis of the verbal report. All participants were undergraduate or graduate students at School of Science and Technology, Meiji University, and had normal vision and hearing. They were paid a small fee for participating in this experiment. Written informed consent was obtained from all participants. The experiment was approved by the ethics committee of the School of Science and Technology, Meiji University, and was conducted in accordance with the principles and guidelines of the Declaration of Helsinki.

Experimental environment

A television monitor 47LW5700-JA (47″ display size; screen inside diameter: 58.6 cm height × 104.3 cm width) made by LG was used for the experiment. Participants’ seating consisted of swivel chairs with arms, placed so that the central axis of their legs was above a line 140 cm away from the front of the TV screen. This allowed a distance of ∼140 cm between the screen and the eyes of each participant. The experiment was performed in a laboratory with the lights dimmed.

Material

The material of the experiment was a Japanese film entitled Dolls (directed by Takeshi Kitano, 2002) depicting Matsumoto (Hidetoshi Nishijima), who rejects his engagement with his fiancée, Sawako (Miho Kanno). In despair, she loses her sanity and ends up in a doll-like state, losing her mind and ability to speak. Seeing her suffering, Matsumoto decides to share her destiny for the rest of his life. The film shows his quiet masculinity and his and Sawako’s strong love for each other together with a sense of emptiness. Since the film’s focus is on these two main characters “wandering together silently”, it appears to provide little information needed to evoke empathy among viewers. However, this extremely limited information is conversely likely to stir the imagination of viewers.

The individual chapters of the film are summarized as follows: Chapter 1 (0:00:00–0:00:44) shows the title and opening credits. Chapter 2 (0:00:45–0:06:34) consists of scenes from “The Courier of Hell” (Meido No Hikyaku), a Bunraku (Japanese puppet drama) play. Chapter 3 (0:06:35–0:08:47) adopts the narrative technique of antichronological order, showing the story’s present. Sawako, who has lost her mind and ability to speak, and Matsumoto, who has decided to share her destiny, are wandering under flowering cherry trees, bound together by a piece of red string. In Chapters 4 and 5, which are back in time, the background of the story is narrated. The main setting of Chapter 4 (0:08:48–0:12:01) is a church where Matsumoto, who has broken up with Sawako, is about to marry the daughter of his company’s president. In Chapter 5 (0:12:02–0:16:21), Matsumoto leaves the church after being informed of Sawako’s suicide attempt and drives to the hospital to which she has been admitted. Chapter 6 (0:16:22–0:21:00) shows a hospital scene where Matsumoto sees Sawako, who appears dissociated.

Perceivers are generally expected to respond to the visual images and story summarized above as follows: In Chapter 3, where the two main characters appear for the first time, perceivers are most likely to be unable to understand the meaning or background of their wandering because of the inverted structure of the chapter. When viewing Chapter 4, which shows friends talking while visiting Matsumoto to attend his wedding ceremony, and scenes from the story’s past, associated with their conversations, perceivers may gradually understand what has happened to the couple. In Chapter 5, where past images of their golden time together, and shots of Sawako losing her sanity are shown with a prolonged shot of Matsumoto displaying a serious facial expression while driving to the hospital, perceivers with growing understanding of the couple’s situation may begin to respond emotionally to it. In Chapter 6, where the couple finally see each other, they both remain speechless and motionless without even making eye contact. There is a prolonged, silent shot of Sawako, who may or may not recognize him, and Matsumoto, who fixes his eyes on her in a peculiar way. In the course of narrative development from Chapter 5 to this scene in Chapter 6, perceivers are expected to have mixed feelings with emotional empathy³⁰ for Sawako while blaming Matsumoto who has unreasonably betrayed her trust in him and cognitive empathy³⁰ evoked while trying to discern the inner feelings of Matsumoto.

NIRS measurement

A multichannel NIRS unit operating at 780-, 805-, and 830-nm wavelengths (OMM-3000, Shimadzu, Kyoto, Japan) was used to measure temporal changes in concentrations of oxy-Hb, deoxy-Hb, and total-Hb. Sixteen optodes were placed on the left and right motor regions in a lattice pattern to form 24 channels for each hemisphere, centered at the C3 and C4 of the EEG 10-20 system (9×9 cm square area). Each channel consisted of one incident optode and one detecting optode located 3 cm apart from the incident optode. The sampling rate was ∼10 Hz. An extensive description of NIRS theory can be found elsewhere.²³

Procedure

Participants were seated and then viewed the opening of Dolls (0:00:00–0:21:00; Chapter 1 to the middle of Chapter 6) on the screen in a natural setting while their brain activity was measured using NIRS. The experiment was done individually. The NIRS data obtained during Chapter 3 (0:06:35∼) and subsequent chapters, in which the two main characters appeared, were analyzed. After viewing the film, participants were instructed to clasp and unclasp their handset about 1–2 Hz (hand motion condition for 5 seconds) or tap their feet while sitting on the chair (leg motion condition for 5 seconds) so that we could identify activated regions of interest (hand and leg motor region in the motor cortex) in the present study. Their brain activity during these motions was also measured. According to the experiment protocol, a 20-second rest was allowed to settle their brain activity before starting each motion, and a set consisting of a 5-second pretask rest, 5-second (hand or leg) motor task, and 10-second post-task rest was repeated six times. A 60-second rest was provided between a hand motion session and a leg motion session.

Data analyses

The film was analyzed each second and flagged according to whether it included any of the following ten regressors to create a design matrix: 1) main male character, 2) main female character, 3) two main characters, 4) other characters, 5) dialogue, 6) switch of shots, 7) switch to the past or the future and omission of time in the present tense, 8) motion of camera, 9) motion of objects, and 10) slow motion. The regressors of interests were 1, 2, and 3, and the regressors of no interests were 4–10. Note that we flagged regressor 3 (two main characters) only when both male and female main characters appeared on the screen at the same time, and unflagged regressors 1 and 2. NIRS data obtained during film viewing were analyzed for each participant using a general linear model, and a group test was performed for each chapter in terms of the three regressors of interest to examine whether there was any significant activation for each channel. Similarly, NIRS data obtained during the hand and leg motion sessions were also analyzed for each channel.

Participants were asked to complete the Interpersonal Reactivity Index (IRI) as well as the questionnaire survey with regard to Dolls after NIRS data acquisition. In Davis’ IRI, conventional emotional empathy scale that cognitive components were downplayed was reexamined and reconstituted into four dimensions by both cognitive and emotional components.^31,32 Perspective-taking measures the tendency to spontaneously adopt the psychological point of view of others. Fantasy measures respondents’ tendencies to transport themselves imaginatively into the feelings and actions of fictional characters. Empathic concern assesses feelings of compassion and concern for unfortunate others. Personal distress assesses “self-oriented” feelings of personal anxiety and unease in response to others’ tense experiences.Perspective-taking and fantasy were designed to measure the cognitive components of empathy. Empathic concern and personal distress were designed to assess the emotional components of empathy. Higher scores of each dimension represent higher empathic tendencies. IRI is frequently used as a measure for social cognitive abilities or empathy in functional neuroimaging literature.^33–35

Results

From the analysis of data obtained during the hand and leg motion sessions, five channels of left hemisphere (ch-1, 2, 5, 13, 15) and three channels of right hemisphere (ch-26, 30, 34) were identified as motor cortices (PFigu0.05,a>