The neural substrate of the ideomotor principle revisited: Evidence for asymmetries in action-effect learning
Highlights
► Investigation of brain activations underlying ideomotor (IM) processing by fMRI. ► In replication, IM response activation was increased for left-hand action effects. ► This hand asymmetry could be linked to an analogous asymmetry in IM learning. ► First description of the neural substrate of IM learning in the literature. ► Caudate nucleus and angular gyrus appeared as central structures of IM learning.
Introduction
The term ideo-motor was coined in the middle of the 19th century, at a time when Europe was captivated by alleged paranormal phenomena ascribed to transcendent powers like table turning or magical pendulums (Tischner, 1929). Carpenter (1882) tried to explain these phenomena by referring to unwilled and unconscious motor excitation elicited by the anticipatory imagination (“idea”) of a specific effect. For instance, thinking of a swinging or rotating pendulum may unconsciously trigger tiny muscle activation in the fingers which hold the pendulum and thereby produce the imagined motion: the “ideomotor reflex”.
Since then, the principle of triggering motor actions by effect anticipations has been embedded into a broad conceptual framework. Today, it is no longer seen as an involuntary reflex, bound to conditions of reduced will and expectant attention, but rather as a ubiquitous mechanism in voluntary action control – a truly executive function (James, 1890; see also Hommel et al., 2001, Pfister and Janczyk, 2012, Shin et al., 2010). In the following, we will refer to the mechanisms that relay sensory anticipations to motor centers as ideomotor response activation (IRA). This process relies on bidirectional associations between motor codes and sensory effect codes that have to be learned (Elsner and Hommel, 2001, Hoffmann et al., 2009). Once such action-effect associations have been acquired, activating a sensory effect code will automatically spread activation to the associated motor codes.
To put it in a broader theoretical framework, ideomotor assumptions can be related to general models of action control or limb praxis. One of the most influential theories of limb praxis was put forward by Rothi et al., 1991, Rothi et al., 1997, a two-route model which distinguishes between the performances of familiar or meaningful movements on the one hand and unfamiliar or meaningless movements on the other. The former and only these would recruit on the so-called “output praxicon”, a specialized long-term mnemonic structure which stores visuo-kinaesthetic attributes of movements, i.e. performance-related sensational or perceptual codes, which for movement execution are directly transcoded into motor programs. Entries of the output praxicon, in turn, get activated by more “passive” perceptual representations of physical characteristics (amplitude, spatial orientation, etc.) of actions, posited to be stored in another neurocognitive structure, the so-called “input praxicon”. According to the theory, the input praxicon allows to identify familiar actions of the agent’s repertoire, whereas the output praxicon supplies the motor implementation of actions at the innervatory pattern stage. Importantly, the outlined executive mechanism avoids the costs incurring for unfamiliar actions which require computing all the parameters needed to implement the spatial and temporal characteristics of intended movements (cf. Rothi and Heilman, 1996). There is an obvious similarity between the theoretically posited functionalities of output praxicon content on the one hand and learned action effects on the other, which are both assumed to be automatically transcoded into motor programs.
Similarly, ideomotor learning can be conceptualized as acquisition of a so-called inverse internal model (Wolpert and Kawato, 1998) which is a feedforward controller of motor action in which the output is identical to the input information. Basically, skillful coordinated limb movements arguably cannot be executed solely under feedback control, because feedback loops are generally slow and have small gains. Therefore, the brain needs to acquire an inverse dynamics model of intended action through motor learning, after which motor control can be executed in a pure feedforward manner (cf. Kawato, 1999, Wolpert and Ghahramani, 2000).
Whereas first neurophysiological studies have targeted the process of IRA (Elsner et al., 2002, Melcher et al., 2008, Kühn et al., 2011), the neural mechanisms underlying the preceding ideomotor learning are virtually unknown. Accordingly, the present study investigated the neural mechanisms underlying ideomotor learning and their relation to subsequent IRA.
To this end, we adopted a two-phase design that was previously used to assess the neurophysiological basis of IRAs (Melcher et al., 2008; cf. also Elsner and Hommel, 2001, Pfister et al., 2011). In an acquisition phase, participants performed key press actions to produce arbitrary action effects which in different subject groups were either contingent or non-contingent with the selected response. Thus, both groups had overall comparable sensory and motor activities but a different potential to exhibit ideomotor learning. In the subsequent test phase, participants of the contingency group were probed for IRA. Effect stimuli (i.e. stimuli which were presented as action effects during the acquisition phase) were now presented together with an imperative target stimulus1, which prompted participants either to freely choose a response or to withhold responding (Fig. 1). No-go trials of the latter kind allow defining the neural correlates of the perception of learned action effects independent of proper motor activation: the pure neural substrate of IRA (cf. Elsner et al., 2002, Melcher et al., 2008). The presence of go trials on the other hand increases the response readiness of subjects and thus assumably promotes effects of IRA during no-go trials2.
As outlined above, previous neurophysiological studies only investigated IRA (in more technical terms: the test phase) and neglected the underlying learning process (the acquisition phase). In these studies, IRA was mirrored in activity of the supplementary motor area (SMA) and the hippocampal system (Elsner et al., 2002, Melcher et al., 2008). The major goal of the present study was to investigate the learning process enabling such response activation effects. Interestingly, response activation effects in previous studies were entirely driven by structures associated with declarative memory such as the hippocampus or the parahippocampal gyrus. This medial temporal memory system is typically distinguished from a second, ‘habit learning system’ in the basal ganglia, i.e. comprising the putamen and caudate nucleus (e.g., Knowlton et al., 1996, Packard and Knowlton, 2002). Given that this second memory system was repeatedly associated with motor learning (see Seger, 2006, for a review), we expected ideomotor learning to draw on this system in addition to the medial temporal system (Tricomi et al., 2004).
Moreover, previous studies suggest that memory-based sensorimotor transformation or integration – i.e. output praxicon function (see above) – is represented in temporo-parietal regions. In this context, Peigneux et al. (2004), for instance, emphasized the contribution of the superior temporal cortex (superior temporal sulcus) in the sensory processing of action-related stimuli or proper motions. Rumiati et al. (2005) report a left-hemispherical pattern of increased activity comprising the inferior temporal gyrus and angular gyrus specifically in response to familiar actions, while Grèzes et al. (1999) related the inferior parietal cortex and the frontopolar cortex (FPC) to the acquisition of familiar actions during action observation (i.e. during visuomotor learning). Based on the outlined findings, ideomotor learning as a special instance of sensorimotor integration can be reasonably expected to rely on temporo-parietal regions in addition to genuine memory- or learning-related structures of the basal ganglia and the hippocampal system.
Furthermore, it is important to note that the described network for IRA found in previous studies only emerged for the left-hand but not for right-hand action-effects, indicating a fundamental asymmetry of ideomotor processes (cf. Melcher et al., 2008). Because the latter conclusion is based on ad hoc findings, it requires further empirical substantiation as well as theoretical elaboration and embedment. In this context, the Action-Perception model (Goodale and Milner, 1992, Milner and Goodale, 2010) emphasizes the differential skillfulness of left-hand and right-hand actions (which is based on the agent’s handedness) as a factor which could mediate their differential proneness to ideomotor processes. The rationale behind is that motor actions which are not highly skilled are controlled by the ventral stream in the same way as perception, whereas highly skilled or automated actions are controlled by the dorsal stream. This functional neuroanatomical similarity of motor and sensory processes assumably facilitates sensorimotor integration including ideomotor processes (cf. Wiediger and Fournier, 2008). However, if motor codes of less skillful left-hand actions are indeed more easily bound to sensory codes of perceived stimuli, this should concern not only IRA but ideomotor learning too. Accordingly, in the present fMRI study, we basically aimed at (1) replicating results of asymmetric IRA in a slightly modified design and (2) to test for analogous asymmetries in the underlying ideomotor learning process. More specifically, based on the results of our previous investigation, we expected to observe increased IRA-related brain activations for the left-hand compared to right-hand learned action effects particularly in medial temporal structures as well as in premotor and supplementary motor cortices. An analogous left-/right-hand side asymmetry for ideomotor learning was expected to occur likewise in activations of medial temporal structures as well as in activations of the basal ganglia.
Section snippets
Participants
Thirty-six healthy, right-handed participants were recruited from the local university’s student community. All subjects gave written-informed consent. They reported normal or corrected-to-normal vision, no history of psychiatric or neurological illness, and were currently not under psychotropic medication. Subjects were randomly assigned to one of the two experimental groups: 20 to the contingency group and 16 to the non-contingency group. Three subjects, however, had to be excluded from
Acquisition phase: neural substrate of action-effect association
Contrasts of the learning regressors are listed in Table 1 and particularly for left-hand action effects visualized in Fig. 2. In a nutshell, we found stronger activation decreases for contingent than for non-contingent action effects for both hands, but we did not observe any stronger activation increases. Contingent right-hand action effects showed a stronger decrease of brain activation in the rostral (perigenual) anterior cingulate cortex, reaching into the corpus callosum (t = 4.51;
Discussion
The present study investigated neural mechanisms that underlie ideomotor processing, particularly the acquisition of action-effect associations by ideomotor learning and the later activation of motor tendencies by the learned action effects (cf. Elsner et al., 2002, Melcher et al., 2008, Kühn et al., 2011).
To our knowledge, this is the first study to investigate the neural substrate of bidirectional action-effect learning. For this purpose, we used a time-by-condition interaction analysis that
Conclusions
In the present work, we investigated the neural mechanisms underlying ideomotor processing, i.e. the acquisition of action-effect associations (ideomotor learning) and the triggering of motor activation by the perception of learned action effects (IRA). We were able to replicate earlier findings of a hand asymmetry in IRA (significantly stronger for left-hand compared to right-hand learned action effects) which we traced back to an analogous hand asymmetry in ideomotor learning. Crucially, to
Acknowledgments
We thank Mr. Timo M.D. Graen (M. Sc. in Physics) for his highly competent support in the programming of the experimental stimulation. Moreover, we are grateful to two anonymous reviewers for their helpful comments and suggestions.
References (61)
Spatial intention-response compatibility
Acta Psychol
(2002)- et al.
Characterizing stimulus–response functions using nonlinear regressors in parametric fMRI experiments
NeuroImage
(1998) - et al.
An fMRI study of reward-related probability learning
NeuroImage
(2005) - et al.
Linking actions and their perceivable consequences in the human brain
NeuroImage
(2002) - et al.
Experiencing oneself vs another person as being the cause of an action: the neural correlates of the experience of agency
NeuroImage
(2002) - et al.
Modulating the sense of agency: a PET study
NeuroImage
(2003) - et al.
Empathy and the somatotopic auditory mirror system in humans
Curr Biol
(2006) - et al.
Separate visual pathways for perception and action
Trends Neurosci
(1992) Internal models for motor control and trajectory planning
Curr Opin Neurobiol
(1999)- et al.
The rewarding value of good motor performance in the context of monetary incentives
Neuropsychologia
(2012)
The neural substrate of the ideomotor principle: an event-related fMRI analysis
NeuroImage
Adaptive control of ideomotor effect anticipations
Acta Psychol
Learning arbitrary visuomotor associations: temporal dynamic of brain activity
NeuroImage
Feedback signals in the caudate reflect goal achievement on a declarative memory task
NeuroImage
Modulation of caudate activity by action contingency
Neuron
Interaction between the human hippocampus and the caudate nucleus during route recognition
Neuron
Multiple parallel memory systems in the brain of the rat
Neurobiol Learn Mem
Multiple paired forward and inverse models for motor control
Neural Netw
Tactile-vibration-activated foci in insular and parietal-opercular cortex studied with positron emission tomography: mapping the second somatosensory area in humans
Somatosens Mot Res
Loss of lever press-related firing of rat striatal forelimb neurons after repeated sessions in a lever pressing task
J Neurosci
The microgenesis of action-effect binding
Psychol Res
Blockade of endogenous opioid neurotransmission enhances acquisition of conditioned fear in humans
J Neurosci
Effect anticipation and action control
J Exp Psychol Human
The angular gyrus computes action awareness representations
Cereb Cortex
Multiple nonprimary motor areas in the human cortex
J Neurophysiol
Improved assessment of significant change in functional magnetic resonance imaging (fMRI): use of a cluster size threshold
Magn Reson Med
Motor sequence learning with the nondominant left hand. A PET functional imaging study
Exp Brain Res
The effects of learning and intention on the neural network involved in the perception of meaningless actions
Brain
Movements or targets: what makes an action in action-effect learning?
Q J Exp Psychol
Cited by (32)
Ideomotor learning: Time to generalize a longstanding principle
2022, Neuroscience and Biobehavioral ReviewsThe neural basis of integrating pre- and post-response information for goal-directed actions
2016, NeuropsychologiaCitation Excerpt :Deviating from this general trend in the neuroimaging literature, two recent fMRI studies investigated the neural correlates of R–O contingency learning while differential outcomes did not concurrently serve as feedback during S–R learning and outcomes were intrinsically non-incentive. In one study by Melcher et al. (2013) subjects were assigned to two different experimental groups and encountered either a contingent or a random mapping between two responses and two non-incentive sound effects. Responses were freely chosen and hence R–O contingencies were not additionally contingent upon an antecedent stimulus.
Functional characteristics of control adaptation in intermodal sensory processing
2015, Brain and CognitionCitation Excerpt :In the same vein, many other studies failed to observe conflict adaptation when the investigated sequenced conflict conditions required different attentional foci (cf. Egner et al., 2007; Funes et al., 2010a; Funes et al., 2010b; Schmidt & De Houwer, 2011; Wendt, Kluwe, & Peters, 2006). This strongly suggests that across-task adaptation effects are ultimately based on, how Notebaert and Verguts (2008) call it, “locally” increased sensory selection rather than increased motor selection (i.e., the shielding of the/an required motor response against competing response tendencies), even though the applied tasks in their study included competition between motor responses (i.e., motor-based conflict) based on stimulus-response overlap (Egner et al., 2007; Kornblum, 1992; Melcher et al., 2013; for a recent exception in terms of transfer from adaptation to emotional interrupts to the processing of Simon interference, see Wirth, Pfister, & Kunde, 2015). In the present study, we extended the approach of previous work on between-task transferability of control adaptation by adopting the reverse perspective: we investigated transferability of control adaptation across different kinds of competition of the same task, rather than transferability across different tasks including the same kind of competition.
Neural correlates of ideomotor effect anticipations
2014, NeuroscienceCitation Excerpt :These results are the first neurophysiological evidence for the brain regions involved in the transformation of anticipated action effects into overt action. The suggested prominent role of the inferior parietal cortex is consistent with recent accounts of parietal function (Sirigu et al., 2004; Fogassi et al., 2005; Goldenberg, 2009; see Melcher et al., 2013, for a role of a related part of inferior parietal cortex – the angular gyrus – for ideomotor learning). For instance, the inferior parietal cortex has been shown to be directly related to the generation of motor intentions in a direct cortical stimulation study (Desmurget et al., 2009).