Traditionally, prehension has been understood as the act of coordinated reaching and grasping. The reaching component of prehension is concerned with bringing the hand to the object to be grasped, whereas the grasping component refers to the opening and closing of the hand. This suggested division of labor stems from the seminal studies performed by Marc Jeannerod, about 25 years ago, which, for the first time, reported details of the kinematics of prehensile movements (Jeannerod 1981, 1984). Countless studies have addressed all kinds of aspects of prehension, taking this division in components as their starting point. Recently, however, Smeets and Brenner (1999) started advocating an alternative to this view of prehension. They suggested that it is not reaching and grasping that make up prehension, but that the individual digits move independently to their respective sides of the object to be grasped, and that what looks like a grasping component is really something emerging from individual digits’ trajectories. The purpose of the study presented here was to critically test Smeets and Brenner’s “double-pointing hypothesis”. Before presenting the experiment, however, we will briefly review the proposals originally made by Jeannerod and the complaints that Smeets and Brenner formulated regarding the traditional division of prehension into a reaching and a grasping component.
As mentioned earlier, with his first systematic analysis of the kinematics of prehension, Jeannerod (1981, 1984) set the stage for a large number of studies of the control and coordination of reaching and grasping (for reviews, see Castiello 2005; Jeannerod 1988; MacKenzie and Iberall 1994). Jeannerod’s proposal that a reaching (or transport) component and a grasping (or manipulation) component make up prehension was based on a number of arguments, such as anatomical arguments that different muscles and brain areas are involved in the control of reaching and grasping (e.g., see Jeannerod 1999; Jeannerod et al. 1995). One of the most prominent among the arguments, however, was that the two components would rely on different types of information about the object: Jeannerod’s “visuo-motor channels hypothesis” (Jeannerod 1981, 1988, 1999; Paulignan and Jeannerod 1996) held that the reaching component operates exclusively on information about extrinsic properties of the object (such as its egocentric distance and direction) and the grasping component operates exclusively on information about intrinsic object properties (such as its size, shape, and surface properties). In this sense, the two components (i.e., the two visuo-motor channels) were hypothesized to be independent. This is why many studies designed to test the independence of the two components of prehension involved perturbations of intrinsic object properties, such as size (e.g., Castiello et al. 1993; Paulignan et al. 1991a) or of extrinsic object properties, such as location (e.g., Gentilucci et al. 1992; Paulignan et al. 1991b), to see if the perturbations would have an effect on the component that should not be dependent on information of either object property. Object-size perturbations, for instance, should only affect the grasping component in Jeannerod’s model.
With two components making up one act (that of prehension), not only their independence but also their coordination becomes an issue. Most often, when the coordination of prehension has been addressed, hypotheses regarding the moment of peak hand aperture, the moment that hand opening goes into hand closing, have been put forward (for an overview, see Zaal and Bootsma 2004). For instance, it has been proposed that peak hand aperture would occur at the moment of peak deceleration of the reaching movement (Jeannerod 1984), at a fixed time (Gentilucci et al. 1992) or distance (Rand and Stelmach 2005; Rand et al. 2006; Wang and Stelmach 1998, 2001) before hand-object contact, or that coordination is based on time-to-contact information (Bootsma and Van Wieringen 1992; Zaal and Bootsma 2004; Zaal et al. 1998). In our opinion, the latter hypothesis is the most promising of the ones currently available, but it certainly still needs a critical test. To do so, however, the conceptualization of prehension into a reaching and grasping component must be valid. This is where Smeets and Brenner’s (1999) hypothesis that prehension should be seen as the combination of independent digit’s movements rather than the combination of reaching and grasping becomes problematic. If prehension is not about reaching and grasping, formulating hypotheses about their coordination (or independence) is pointless. This was the direct inspiration for the current study. But before we turn to the experiment that we performed, let us see what made Smeets and Brenner propose a new view on prehension.
Smeets and Brenner (1999) formulated a number of points of dissatisfaction with the original division of labor between a grasping and reaching component as initially proposed by Jeannerod (1981). Smeets and Brenner pointed out that the distinction between intrinsic and extrinsic object properties was problematic. For instance, the orientation of an object could be (and has been) seen both as an intrinsic and as an extrinsic object property. To identify the respective visuo-motor channels on the basis of their exclusive reliance on information about these two types of object properties, Smeets and Brenner argued, was impossible. Furthermore, Smeets and Brenner explained that the anatomical arguments for the distinction of reaching and grasping components were invalid as well. To argue that reaching (hand transport) and grasping (shaping the hand) rely on to the use of proximal and distal muscles, respectively, an argument used by Jeannerod to distinguish the two components of prehension, did not convince Smeets and Brenner, who pointed at the fact that, for instance, polyarticular muscles in the lower arm (which are proximal muscles) are involved in movements of the digits.
As an answer to what they called the “classical approach” of Jeannerod (1981), Smeets and Brenner (1999) presented an “alternative approach”. Their approach essentially proposes to think of prehension as the independent movement of the contributing digitsFootnote 1 to their respective planned end positions. These digits, as Smeets and Brenner argued, typically arrive at the surface more or less perpendicularly. If one would look at the average path of the two digits, this would be the straight path that might look like a reaching movement; looking at the distance between the two independently moving digits as a function of time would show the well-known hand-aperture profile (and might be incorrectly interpreted as such, according to Smeets and Brenner). To demonstrate how this control of independent digits looking like reaching and grasping might work, Smeets and Brenner modeled the kinematics of the individual digits with the minimal jerk model (Flash and Hogan 1985), but now with a non-zero deceleration at the moment of digit-hand contact. This latter, final deceleration, scaled by movement time squared, made up an “approach parameter”.
Smeets and Brenner demonstrated in their original study as well as in others (e.g., Smeets and Brenner 1999, 2001; Smeets et al. 2002) how by varying the approach parameter and movement time the model fitted empirical data. To appreciate the close resemblance of the model behavior with experimentally established relations among “intrinsic” and “extrinsic” object properties and the kinematics of prehension, Smeets and Brenner invited us to translate the average trajectory of the thumb and index finger into a hand transport trajectory and the difference of the trajectories of the thumb and index finger into a grasping trajectory, of course, only for the purpose of comparing the model to observed kinematics. Simulations of the model showed that reaching is not affected by variations in “intrinsic” object properties, that grasping component is not affected by variations in “extrinsic” object properties, that peak hand aperture occurs later in the movement for larger objects, and that an increase in the approach parameter, for instance because a slippery object surface asks for a more perpendicular approach of the digits, leads to a larger peak hand aperture occurring relatively earlier in the movement.
As we discussed earlier, if the hypothesis of Smeets and Brenner’ (1999) is true that it is the digits themselves which are controlled in prehension and not a reaching and grasping component, much of the research on prehension, most notably the studies on the independence of the reaching and grasping component and the studies on the coordination of the two putative components, have been pointless. That is why, we think, a well-funded appraisal of either Smeets and Brenner’s “new view” or of the “classical approach” is called for. Although there have been a number of theoretical and methodological arguments against Smeets and Brenner’s new view (e.g., Marteniuk and Bertram 1999; Newell and Cesari 1999; Rosenbaum et al. 1999; Steenbergen 1999), we felt that an empirical test would be the strongest argument in favor of either approach. This is the reason why we set out to test Smeets and Brenner’s account of prehension. The logic behind our test is the following. If Smeets and Brenner have been correct with their hypothesis that the two digits that are used to pick up an object with a precision grip (between thumb and index finger) move independently to their respective end positions on the object, changing the end position of one of the digits, say, the thumb, would not have an effect on how the other digit, in this case the index finger, would move to its, unchanged, end position. In other words, changing the end position of one digit during the movement should not affect the kinematics of the other digit. For the experiment, we developed an object of which both side surfaces could be made to quickly slide in or out independently (see the Supplementary Movie). We had participants reach for and grasp this object. In some trials, we had one of the two side surfaces slide in or out right after the movement had started, such that one digit had to move to a new position whereas for the other nothing had changed. By comparing the kinematics of both digits with that of unperturbed trials, we were able to test Smeets and Brenner’s hypothesis.