Harleß’ Apparatus of Will: 150 years later

Over the last two decades, ideomotor theory as a 19th century philosophical approach to human action has stimulated numerous intriguing experiments on this fundamental mechanism (for recent reviews see Nattkemper, Ziessler, & Frensch, 2010; Shin, Proctor, & Capaldi, 2010). While these experiments steadily increased our understanding of the ideomotor mechanism, classical comprehensive accounts of ideomotor action control are about to fall into oblivion. Symptomatically, of the several hundred papers on ideomotor theory published to date, only one targeted earlier formulations of the theory (Stock & Stock, 2004).¹

Among these earlier formulations are several comprehensive accounts for ideomotor theory, starting with Herbart’s (1825) Psychologie als Wissenschaft neu gegründet auf Erfahrung, Metaphysik und Mathematik, and continuing with Lotze’s (1852) Medicinische Psychologie oder Physiologie der Seele and Harleß’ (1861) Der Apparat des Willens. These classical works offer a wealth of intriguing ideas and point towards important questions that contemporary research has not answered yet. Still, they are not as well known as would be expected based on their potential merit for contemporary research. At least two factors are responsible for this circumstance: First, most classical treatises clearly differ from contemporary research papers in that their length exceeds the concise format of the latter ones where relevant information is assumed to be more readily extractable from a minimum of pages. The second factor is obvious from the titles cited above: most comprehensive 19th century accounts of ideomotor theory were written in German language and are thus inaccessible to the majority of contemporary researchers. The present article aims at making the most concise of these classical accounts (Harleß, 1861) available to the English-speaking world. To this end, the following sections first provide a few biographical notes on the author and continue with a summary of the key points in his Der Apparat des Willens.

In contrast to his philosophical predecessors Herbart and Lotze, Emil Harleß (Fig. 1) was a physiologist, who had studied medicine, physics, and chemistry in Würzburg and Berlin (Meyers Konversationslexikon 1885-1892). During the first years of his academic career at the University of Erlangen, he focused on experiments on general anaesthetic drugs and published a seminal report in this area together with Ernst von Bibra (Bibra & Harleß, 1847).

Only after becoming a full professor at the University of Munich (Martin, 1857) he started to investigate the “physiological-psychological mechanism”, what would today be subsumed under the terms physiological psychology or cognitive neuroscience (e.g., Harleß, 1862). One year before his death—exactly 150 years ago—he published an article where he discussed ideomotor theory in the broader framework of the mind–body problem. This article—Der Apparat des Willens [The Apparatus of Will]—is arguably the most concise and comprehensive account for ideomotor action formulated prior to James’s (1890/1981) Principles of Psychology (cf. Hommel, 2003; Stock & Stock, 2004). It contains a wealth of innovative ideas, some of which have already received empirical support and others that are just now being addressed by experimental studies. An English translation of the Apparatus of Will is available as supplementary online material whereas the following sections highlight its most important ideas for contemporary research.²

Contemporary accounts of ideomotor theory concentrate on the immediate role of sensory anticipations for the production of motor actions (Hoffmann, 1993, 2003; Hommel, 2009; Hommel, Müsseler, Aschersleben, & Prinz, 2001; Kunde, 2001, 2003). To the contrary, Harleß (1861) conceives ideomotor theory as a broader approach to the philosophical mind–body problem, including the classical controversy about human free will and related issues such as its legal implications. Consequently, the first part of the article (“The foundations of voluntary action”) contains a thorough discussion of the relationship of voluntariness, consciousness, and purposefulness of voluntary actions.

Regarding this distinction, Harleß (1861) arrives at the interesting conclusion that an action is most likely to appear voluntary to an observer if it does not serve an apparent purpose, i.e., when it does not seem to be motivated by situational demands (labelled incidental actions). The conclusion that such incidental actions give rise to the most intense impression of voluntariness is supported by several recent studies on imitation behaviour. In these studies, human infants and chimpanzees tended to imitate unusual object-oriented actions, such as pressing a button with one’s forehead, only if the agent could have used his hands easily instead of the forehead but not if the agent’s hands were occupied (e.g., Buttelmann, Carpenter, Call, & Tomasello, 2007; Gergely, Bekkering, & Király, 2002). In accordance with Harleß’ deduction, actions that do not seem to be motivated by situational demands (in this case: having his hands free instead of carrying an object), do indeed seem to have a special influence on the observer. The discussion of such incidental actions also leads to a fine-grained analysis of voluntary action. For instance, Harleß suggested that voluntary actions do comprise decisions about whether an action should be carried out at all—a notion that was recently taken up in the What-When-Whether model of intentional action (Brass & Haggard, 2008).

The first part of the article ends with the claim that incidental actions are the key to understanding voluntary actions in general. This idea was addressed in various recent studies where the term ‘incidental actions’ was substituted by the concepts of self-initiated (Jahanshahi et al., 1995), endogenous (Astor-Jack & Haggard, 2005), and intention-based actions (Herwig, Prinz, & Waszak, 2007; Pfister, Kiesel, & Melcher, 2010). The question of how such actions—in contrast to immediate reactions to external stimuli—might promote ideomotor learning or the anticipation of distal action effects is just now being addressed by empirical studies (Dutzi & Hommel, 2009; Haering & Kiesel, (2011); Herwig et al., 2007; Herwig & Waszak, 2009; Janczyk, Pfister, & Kunde, (2011); Pfister et al., 2010; Pfister, Kiesel, & Hoffmann, 2011; Stock & Hoffmann, 2002; Wolfensteller & Ruge, (2011)).

The second part of the article (“The physiology of incidental actions”) is concerned with the physiological implementation of voluntary actions in general and incidental actions in particular. Like other physiologists of his time (e.g., Laycock, 1845), Harleß (1861) assumes two qualitatively different functional units of the nervous system: a sensory unit (sensorium) and a motor unit (motorium). Accordingly, activations have to be relayed from the former system to the latter system in order to produce adaptive action. A considerable proportion of the article is devoted to this question which has had a relatively limited impact on contemporary research. Notable exceptions include recent neuroimaging studies on the acquisition of bidirectional action-effect associations (Elsner et al., 2002; Melcher, Weidema, Eenshuistra, Hommel, & Gruber, 2008), the sensory nature of effect anticipations (Kühn, Seurinck, Fias, & Waszak, 2010; Kühn, Keizer, Rombouts, & Hommel, 2011), and the electrophysiological signature of effect anticipations (Waszak & Herwig, 2007; Nikolaev, Ziessler, van Leeuwen, & Dimova, 2008). In this respect, Harleß seems to have foreseen the advent of neuroimaging methods in claiming that “activity of the soul” will inevitably increase the blood flow in the activated brain regions—preparing the theoretical stage for an fMRI analysis of ideomotor processes. Still, however, the neurophysiological mechanisms that relay effect anticipations to motor centres are virtually unknown and will have to be uncovered in the years to come. A promising start in this enterprise might be the integration of ideomotor theory and recent concepts in computational neuroscience, such as forward and inverse models (e.g., Wolpert & Kawato, 1998).

A third part of Harleß’ (1861) article covers the acquisition of bidirectional action-effect representations for humans and animals (“The acquisition of behavioural competence”). Like prior accounts (Lotze, 1852), this description is relatively vague; yet, it suggests an important role of incidental movements during prenatal development. This discussion already contains important thoughts about the contextual nature of action-effect representations that were later addressed empirically (Hoffmann & Sebald, 2000; Kiesel & Hoffmann, 2004). The later ontogenetic development of action-effect representations is not covered in the Apparatus of Will but is certainly an important addition to the considerations raised by Harleß (e.g., Eenshuistra, Weidema, & Hommel, 2004; Karbach, Kray, & Hommel, 2011; Verschoor, Weidema, Biro, & Hommel, 2010). Interestingly, Harleß explicitly suggested the acquisition of action-effect representations to be subjected to preparedness for learning specific action-effect relations while other relations are much harder to acquire. Even though this notion was empirically tested in operant conditioning experiments with rodents (Bolles, 1970; Randolph, 1986) it has not yet been applied to human agents.³

The major part of Harleß’ (1861) article describes a physiologically motivated schematic of action control via ideomotor effect anticipations (“A physiological-psychological mechanism for voluntary actions”; Fig. 2). Interestingly, Harleß already employed the modern term of action effects and pointed towards the contextual dependence of action-effect associations (Hoffmann, 1993; Hoffmann & Sebald, 2000). In contrast to contemporary experiments which exclusively studied the situational context, Harleß proposed a much broader definition, including the internal context such as the current mood or arousal level. Whereas the internal context has been investigated in other domains of psychology (e.g., regarding mood-congruent memory; Bower, 1981), its role for ideomotor processes still has to be uncovered. Furthermore, Harleß explicitly discusses the role of proximal and different types of distal effects—a discussion that empirical studies are just now beginning to take up (Janczyk, Skirde, Weigelt, & Kunde, 2009). Importantly, Harleß also emphasises a unique role of the valence of actions and their effects for acquisition and application of action-effect associations that, up to now, has only played a minor role for contemporary research and theory (but see Beckers, De Houwer, & Eelen, 2002; Eder, Müsseler, & Hommel, (2011); Kunde, Lozo, & Neumann, (2011)). Both, the distinction between proximal and distal feedback and the role of valenced action effects, have direct implications for future research. Action effects are no longer to be seen as a homogenous class of events that are only defined by their contingent mapping to an action. To the contrary, the exact properties of action effects have to be taken into account to arrive at a clearer understanding of their role for human and non-human action control.