Elsevier

NeuroImage

Volume 138, September 2016, Pages 109-122
NeuroImage

Violating instructed human agency: An fMRI study on ocular tracking of biological and nonbiological motion stimuli

https://doi.org/10.1016/j.neuroimage.2016.05.043Get rights and content

Highlights

  • Mismatch of agency belief and motion kinematics causes medial–frontal activation.

  • In the oculomotor network, the supplementary eye fields mediate this mismatch.

  • Activation is asymmetric as violating computer agency yields no additional activation.

  • Occipito-temporal areas including V5 show a preference for biological motion.

Abstract

Previous studies have shown that beliefs about the human origin of a stimulus are capable of modulating the coupling of perception and action. Such beliefs can be based on top-down recognition of the identity of an actor or bottom-up observation of the behavior of the stimulus. Instructed human agency has been shown to lead to superior tracking performance of a moving dot as compared to instructed computer agency, especially when the dot followed a biological velocity profile and thus matched the predicted movement, whereas a violation of instructed human agency by a nonbiological dot motion impaired oculomotor tracking (Zwickel et al., 2012). This suggests that the instructed agency biases the selection of predictive models on the movement trajectory of the dot motion. The aim of the present fMRI study was to examine the neural correlates of top-down and bottom-up modulations of perception–action couplings by manipulating the instructed agency (human action vs. computer-generated action) and the observable behavior of the stimulus (biological vs. nonbiological velocity profile). To this end, participants performed an oculomotor tracking task in an MRI environment. Oculomotor tracking activated areas of the eye movement network. A right-hemisphere occipito-temporal cluster comprising the motion-sensitive area V5 showed a preference for the biological as compared to the nonbiological velocity profile. Importantly, a mismatch between instructed human agency and a nonbiological velocity profile primarily activated medial–frontal areas comprising the frontal pole, the paracingulate gyrus, and the anterior cingulate gyrus, as well as the cerebellum and the supplementary eye field as part of the eye movement network. This mismatch effect was specific to the instructed human agency and did not occur in conditions with a mismatch between instructed computer agency and a biological velocity profile. Our results support the hypothesis that humans activate a specific predictive model for biological movements based on their own motor expertise. A violation of this predictive model causes costs as the movement needs to be corrected in accordance with incoming (nonbiological) sensory information.

Introduction

Previous research suggests that the observation of human actions is a special case in the perception of dynamic events in the environment, and radically different from observing nonhuman actions such as object motion or animal behavior (e.g. Wilson, 2001, Wilson and Knoblich, 2005). Accordingly, the presence of a human agent can modulate various aspects of human information processing, such as motion perception (e.g. Shiffrar and Freyd, 1990, Shiffrar and Freyd, 1993), attention (e.g. Frank et al., 2009, Klin et al., 2009), and language comprehension (e.g. Bornkessel-Schlesewsky and Schlesewsky, 2009, Frenzel et al., 2015). Most interestingly for the present study, human agency has also been shown to alter the coupling of perception and action in a variety of experimental paradigms such as stimulus–response-compatibility (e.g., Brass et al., 2000, Brass et al., 2001, Craighero et al., 2002, Press et al., 2005), motor priming (e.g., Castiello et al., 2002, Edwards et al., 2003), or visuomotor coordination (e.g., Bouquet et al., 2007, Stanley et al., 2007, Zwickel et al., 2012; for reviews, see Hegele, 2009, Press, 2011). For example, Kilner et al. (2003) employed a continuous movement execution and observation paradigm demonstrating that the coupling of perception and action is apparently sensitive to the presence of a human agent. More specifically, they found interference between perception and action when the participant's arm movement did not match the simultaneously observed arm movement of a human model, but no interference when the observed movement was made by a robot model. The authors argued that the observed motion of a human model led to an automatic activation of the corresponding motor program in the observer which then interfered with movement execution (see also, Blakemore and Frith, 2005).

With respect to the necessary conditions that could give rise to such specific interference effects, the mere belief about the origin of an observed motion seems to be sufficient to modulate the coupling of perception and action as similar interference effects have been observed for human hand movements as well as abstract dot motions. Using a similar task as Kilner et al. (2003); Stanley et al. (2007) found interference between simultaneous movement observation and execution when participants were instructed beforehand that the subsequent dot motion was generated by a human agent, regardless of whether it followed a biological or a nonbiological velocity profile. If participants had been informed that the dot motion was generated by a computer, interference effects occurred neither for biological nor for nonbiological dot motions. These findings suggest that prior beliefs regarding the origin of an observed movement influence its processing and the degree to which movement observation automatically activates a corresponding motor program that can then either facilitate or interfere with the execution of one's own movement. Accordingly, the authors concluded that top-down influences such as expectations or prior beliefs of human agency are more important factors of interference effects than bottom-up stimulus kinematics. Even though this claim in its strong form seems to be questionable, as it has been shown that both agency and motion profiles impact upon interference effects (Kilner et al., 2007), a subsequent study by Zwickel et al. (2012) provided further support for the importance of top-down assigned human agency on continuous perception–action couplings. They found that oculomotor tracking performance was superior (e.g. less catch-up saccades) when an observed dot motion followed a biological velocity profile as compared to a nonbiological (i.e. constant) velocity profile. Importantly, the tracking performance for biological velocity profiles was modulated by the observer's belief about the origin of the observed movement: the advantage was particularly pronounced when participants were instructed that the observed dot motion was produced by a human agent as compared to being computer-generated. To account for this finding, the authors argued that the central sensorimotor system uses predictive models to anticipate the movement trajectory of the dot to be tracked and that the selection of such predictive models could be biased based on verbal instructions about the origin of an observed movement (cf. Haruno et al., 2001, Imamizu et al., 2007, Wolpert and Kawato, 1998). Whenever a mismatch between the instructed agency (human vs. computer) and the actual velocity profile (biological vs. nonbiological) is detected, costs incur leading to worse tracking performance. Furthermore, Zwickel et al. (2012) argued that this mismatch effect is asymmetric in nature. Only the human agency instruction should activate specific predictive models based on the individual's specific motor expertise in performing movements with biological kinematics that have to be corrected by incoming sensory information about the (nonbiological) dot motion. Accordingly, one could argue that the computer agency instruction should not activate specific predictive models due to (a) the predominance of biological prediction models as a result of one's own motor expertise, and (b) the lack of visual expertise in observing computer-generated motion of single dots on a computer screen.

Top-down modulations of human and non-human agency beliefs on perception have been associated with different brain areas. Stanley et al. (2010) used scrambled versions of point-light animations of human movements (e.g., walking, punching) which ranged from low, medium to high stimulus realism. The stimuli were combined with an instruction of human or computer agency. Participants were asked to provide a perceptual judgment by rating the agreement between the agency category and the perception of the stimulus. They rated physically identical stimuli more often as human when the stimulus was preceded by the instruction human compared to computer and vice versa. The belief of human agency was associated with activation in the ventral paracingulate cortex, the orbital gyrus, and the occipito-temporal regions, while the belief of computer agency activated the ventrolateral prefrontal cortex, parietal and temporal areas as well as the cerebellum. When the observed stimulus did not match the agency instruction (e.g., human agency instruction with low stimulus realism), stronger activation was found in the dorsal paracingulate gyrus, the pre-SMA, the inferior parietal lobule (IPL), and the superior parietal lobule (SPL). The authors emphasized the role of the paracingulate cortex in integrating prior beliefs with bottom-up stimuli and in detecting a mismatch between the two sources of information.

But how does the belief of human agency influence sensorimotor processing on the cortical level? The fMRI study by Stanley et al. (2010) applied a perceptual-judgment task and thus differs from previous behavioral work using arm (Stanley et al., 2007) or eye (Zwickel et al., 2012) movements to examine the processing of human agency beliefs with respect to perception-action couplings. Since differential neural networks are assumed to be recruited depending on whether the task requires a perceptual judgment about a stimulus or an interaction with the stimulus (e.g., Goodale and Westwood, 2004), the aim of the present experiment was to examine the neural correlates of top-down and bottom-up modulations of human agency in a dynamic perception–action coupling task.

Oculomotor tracking is well suited to investigate the interaction of bottom-up stimulus and top-down agency modulations on perception–action coupling as the oculomotor system is sensitive to biological and nonbiological movement kinematics (de'Sperati and Viviani, 1997, Soechting et al., 2010) as well as to human agency instructions (Zwickel et al., 2012). We based our experimental design on the behavioral study by Zwickel et al. (2012) and used the same motion stimuli. Participants were placed in an MRI scanner and asked to track a dot moving along an elliptical path with their eyes. We manipulated the velocity profile of the dot motion (biological vs. nonbiological) and the agency instruction (human- vs. computer-generated) and combined the two, creating either a match (human–biological or computer–nonbiological) or a mismatch (human–nonbiological or computer–biological). Based on the asymmetric mismatch effects demonstrated by Zwickel et al. (2012) we specifically focus on the condition in which instructed human agency is violated by a nonbiological dot motion.

We hypothesize activation in the oculomotor network comprising the frontal eye fields (FEF), the supplementary eye fields (SEF), and posterior parietal areas (Berman et al., 1999) as well as early visual areas, in particular the motion-sensitive area V5 (Burke and Barnes, 2008, Nagel et al., 2006, Nagel et al., 2008, Petit and Haxby, 1999; for a review see Lencer and Trillenberg, 2008), independent of the velocity profile and the agency instruction. Based on the hypothesis of a predominance of biological predictive models and the lack of a meaningful concept of nonbiological dot motion as implied by behavioral findings (cf., Stanley et al., 2007, Zwickel et al., 2012), we expect that a mismatch can only be detected for instructions of human agency and, as a consequence, specifically occurs when the human instruction is combined with a nonbiological dot motion. Thus, activation in this condition should differ from all other conditions. At the cortical level, we hypothesize activation in brain areas previously shown to be involved in the detection of a mismatch between instructed agency and perceived stimulus motion (Stanley et al., 2010). In particular, we expect stronger activation in the paracingulate gyrus, the IPL, and the SPL for violated human agency compared to the remaining conditions (Stanley et al., 2010). In addition, we hypothesize that top-down processing of instructed agency (human vs. computer) is primarily reflected by higher activation in medial and lateral prefrontal regions (Stanley et al., 2010), while bottom-up processing of dot motion (biological vs. nonbiological) primarily activates striate and extrastriate visual areas (Stanley et al., 2010) as well as adjacent temporal areas (Grossman et al., 2000, Pelphrey et al., 2003). Due to the oculomotor component in the present task, we further expect a modulation of brain activity in the eye movement network depending on stimulus velocity (e.g., Nagel et al., 2006, Nagel et al., 2008).

Section snippets

Participants

Twenty-one participants took part in this fMRI study. We excluded one participant because the timing of the experiment was not logged due to technical problems and another participant due to large motion artifacts (see Section 2.6.1) leaving a final sample of 19 participants (mean age = 24.8, range = 20–31 years, 10 female). All participants were right-handed as assessed with the Edinburgh Handedness Inventory (Oldfield, 1971), had normal vision, and no history of neurological or psychiatric

Results

In the present study, we examined the neural correlates of a previously observed modulatory influence of top-down agency beliefs and bottom-up movement kinematics on the coupling of perception and action. To this end, we analyzed brain activation during oculomotor tracking of a simple moving stimulus with varying agency instructions (human vs. computer-generated) and movement velocity profiles (biological vs. nonbiological). First, we examined the activation overlap of all conditions in order

Discussion

In this study we investigated the neural correlates of top-down and bottom-up modulations of human agency on dynamic perception–action coupling by manipulating instructions about the origin of an observed movement (human- vs. computer-generated) as well as movement kinematics (biological vs. nonbiological velocity profile) during an oculomotor tracking task. Across all conditions, oculomotor tracking strongly activated brain areas, previously designated as the eye movement network (Krauzlis,

Conclusions

The present results suggest that areas within the FP, paracingulate cortex, ACC, SEF, and the cerebellum are part of a network processing discrepancies between external sensory information and internal predictions. Most importantly, mismatch-related activation within this network apparently reflects the asymmetrical selection of biological rather than nonbiological predictive models as hypothesized previously based on behavioral work. Thus, our results highlight the special role of (beliefs

Acknowledgments

This work was supported by the German Research Foundation DFG TRR 135 and DFG Fi1567/4-1.

References (84)

  • K.J. Friston et al.

    Event-related fMRI: characterizing differential responses

    NeuroImage

    (1998)
  • M.A. Goodale et al.

    An evolving view of duplex vision: separate but interacting cortical pathways for perception and action

    Curr. Opin. Neurobiol.

    (2004)
  • J. Grèzes et al.

    Inferring false beliefs from the actions of oneself and others: an fMRI study

    NeuroImage

    (2004)
  • E.D. Grossman et al.

    Repetitive TMS over posterior STS disrupts perception of biological motion

    Vis. Res.

    (2005)
  • M. Jenkinson et al.

    Improved optimization for the robust and accurate linear registration and motion correction of brain images

    NeuroImage

    (2002)
  • J. Kilner et al.

    An interference effect of observed biological movement on action

    Curr. Biol.

    (2003)
  • O.E. Krigolson et al.

    Hierarchical error processing: different errors, different systems

    Brain Res.

    (2007)
  • F. Lacquaniti et al.

    The law relating the kinematic and figural aspects of drawing movements

    Acta Psychol.

    (1983)
  • R. Lencer et al.

    Neurophysiology and neuroanatomy of smooth pursuit in humans

    Brain Cogn.

    (2008)
  • M. Nagel et al.

    Parametric modulation of cortical activation during smooth pursuit with and without target blanking. An fMRI study

    NeuroImage

    (2006)
  • M. Nagel et al.

    Cortical mechanisms of retinal and extraretinal smooth pursuit eye movements to different target velocities

    NeuroImage

    (2008)
  • R.C. Oldfield

    The assessment and analysis of handedness: the Edinburgh inventory

    Neuropsychologia

    (1971)
  • N.N. Oosterhof et al.

    Crossmodal and action-specific: neuroimaging the human mirror neuron system

    Trends Cogn. Sci.

    (2013)
  • M.V. Peelen et al.

    Patterns of fMRI activity dissociate overlapping functional brain areas that respond to biological motion

    Neuron

    (2006)
  • C. Press

    Action observation and robotic agents: learning and anthropomorphism

    Neurosci. Biobehav. Rev.

    (2011)
  • C. Press et al.

    Robotic movement elicits automatic imitation

    Cogn. Brain Res.

    (2005)
  • R. Saxe et al.

    A region of right posterior superior temporal sulcus responds to observed intentional actions

    Neuropsychologia

    (2004)
  • S.M. Smith et al.

    Advances in functional and structural MR image analysis and implementation as FSL

    NeuroImage

    (2004)
  • J. Stanley et al.

    How instructions modify perception: an fMRI study investigating brain areas involved in attributing human agency

    NeuroImage

    (2010)
  • D. Wolpert et al.

    Multiple paired forward and inverse models for motor control

    Neural Netw.

    (1998)
  • M.W. Woolrich et al.

    Multilevel linear modelling for FMRI group analysis using Bayesian inference

    NeuroImage

    (2004)
  • Y. Yomogida et al.

    The neural basis of agency: an fMRI study

    NeuroImage

    (2010)
  • E.J. Anderson et al.

    Cortical network for gaze control in humans revealed using multimodal MRI

    Cereb. Cortex

    (2012)
  • L. Avanzino et al.

    The cerebellum predicts the temporal consequences of observed motor acts

    PLoS One

    (2015)
  • J. Barton et al.

    Functional MRI of lateral occipitotemporal cortex during pursuit and motion perception

    Ann. Neurol.

    (1996)
  • R. Berman et al.

    Cortical networks subserving pursuit and saccadic eye movements in humans: an FMRI study

    Hum. Brain Mapp.

    (1999)
  • I. Bornkessel-Schlesewsky et al.

    The role of prominence information in the real time comprehension of transitive constructions: a cross-linguistic approach

    Lang. Linguist. Compass

    (2009)
  • C.A. Bouquet et al.

    Influence of the perception of biological or nonbiological motion on movement execution

    J. Sports Sci.

    (2007)
  • G. Buccino et al.

    Action observation activates premotor and parietal areas in a somatotopic manner: an fMRI study

    Eur. J. Neurosci.

    (2001)
  • M.R. Burke et al.

    Brain and behavior: a task-dependent eye movement study

    Cereb. Cortex

    (2008)
  • U. Castiello et al.

    Observing a Human or a Robotic Hand Grasping an Object: Differential Motor Priming Effects

  • J.D. Connolly et al.

    A comparison of frontoparietal fMRI activation during anti-saccades and anti-pointing

    J. Neurophysiol.

    (2000)
  • 1

    Shared last authorship.

    View full text