3.4. Discussion 126
3.4.1 Priming and goal-set 126
When Gibson first described affordances (1979), he stressed that the kinds of actions an object would elicit would greatly depend on the goal state of the individual; for example, a fallen tree may afford sitting if one is tired, but may also afford standing if one is looking for a better view (Humphreys et al., 2009). In Experiment 1 we show that real grasping is differentially sensitive to priming depending on task goals; when the task
ess simply move an object aside. At a glance, such
terpretations appear at odds ith other studies that have shown effects of object semantics on grasping kinematics (Gentilucci, 2002, 2003; Gentilucci, Benuzzi, Bertolani, Daprati, & Gangitano, 2000; Glover & Dixon, 2002; Glover et al., 2004). However, in most of these studies the effects of semantics were mediated by words, which is different than testing for semantic effects mediated by the sight of objects. Also, in these previous studies semantic effects on grasping were driven by features such as object size or location, not learned function or use. We view our findings as convergent evidence that the functional and metrical aspects of grasping can be dissociated; both priming and the kinematics of grasping were dependent on the task. Previously, Riddoch et al. (Humphreys & Riddoch, 2000; Riddoch et al., 2000) showed that affordance effects in an individual with utilization behaviour depend on task; for the handle of a cup to evoke incorrect responses, the task had to be grasping (see also, Riddoch et al., 1998). These findings coincide nicely with ours, and suggest that task goals shape and determine affordances.
However, immediate task goals were not the only determining factor; when GTM and GTU trials were intermixed (Experiment 3.2), priming emerged for GTM trials as well (Figure 3.5). These findings, although at first surprising to us, are in fact quite consistent with the results of a previous study by Bub et al. (2003). In this work, was GTU, priming was evident, but when the task was GTM, priming was not (Figure 3.2). In line with our a priori predictions, priming effects appeared to follow the relative importance of tool identity to the particular goals and requirements of the task. We interpret these findings as evidence that grasping can proceed without the need to proc object identity when the goal is to
participants were trained to respond to different colours by demonstrating different hand learned object-use interaction style (e.g. a “poke” hand gesture). When participants were
and on others respond to colour
t
ls,
d colleagues (2009), to suggest that the goal-set, or action-state, of an individual indeed helps to
made to distracters which shared the same orientation as the target much more often if gestures, each of which fit with a specific
later cued to name objects on some trials , there were effects in the gesturing-to-colour responses based on whether or not the object in question shared the same hand configuration when used. However, in a separate experiment, when naming trials were not included no such effects were observed. Although not the focus of the paper, this finding was taken to indicate that affordances associated with the use of objects are not automatically evoked, but are instead only evoked when a certain extent of processing is, (or has been), directed to the object itself; in this case, recognition and retrieval of object names on separate trials did the trick. It seems that drawing attention to object identity on naming trials led to a ‘spill over’ of attention to these features on gesture trials. With attention to objec identity in place, use-related affordances were evoked on gesture trials, and influenced response times accordingly. Our priming results for GTM, absent when tasks were blocked but evident when our tasks were put together, are in line with these findings. Attention to tool identity in GTU trials turned attention to these features in GTM tria and, with attention to identity in place, priming effects emerged for GTM trials.
In this view, task goals modulate attention to particular object features, which in turn modulate corresponding affordances (and priming). Thus, affordance effects are determined by task goals insofar as task goals determine the allocation of attentional resources. If we expand this idea, and define the goal-set of an individual as that which is
also determined by overall task setting and context, as well as immediate goals and intentions, then we can see how changes in setting can lead to changes in affordance effects, and this can explain why we observed priming for GTM trials in the mixed but not blocked experiments. In fact, there is a host of other evidence, nicely reviewed by Humphreys an
shape processes of attention and selection. For example, Bekkering and Neggers (2002) asked participants to search for a target defined by its colour and orientation (e.g., find the red horizontal bar). Participants had to indicate the target by either pointing to it, or by grasping it. Eye movements during the search were
participants were required to grasp versus point. It was as if the action-state of the individual influenced the ‘weighting’ of visual features in the display; in the case of grasping, orientation was weighted more strongly, presumably because orientation is more important for grasping than is object colour.
With these views in mind, our findings can be explained by differences in the way attention operates on (e.g. activates) affordances in accordance with the particular goal- set of the individual. Notably, however, our priming method may be limited in its
capacity to detect affordances. For example, perhaps the presentation of a given tool as a prime leads to the activation of associated motor plans in both tasks, regardless of setting (blocked or mixed), but such activations then decay at different rates, or are differentially inhibited. Such differences could then impact priming and account for our results. In patients demonstrating utilization behaviour, the tendency to compulsively grasp and interact with objects is believed to reflect a failure to properly monitor and inhibit environmentally-driven processes (Lhermitte, 1983; Shallice et al., 1989). Lhermitte proposed that visual inputs activate parietal representations for actions, and that normally such activations are held in check by frontal control mechanisms. Damage to key frontal areas “releases the activity of the parietal lobe”, which then “tends to subject the patient to all e
l y,
as to xternal stimuli”, and utilization behaviour results. Our findings may reflect differences in the extent of such inhibitory control depending on task goals and overal task setting. When GTM trials were blocked, suppression of affordance-driven activit specific to specific tools, may have allowed for more efficient responses. In contrast, in the context of the mixed design it may have been best to ‘allow’ such activations to unfold for both tasks. Switching inhibition of affordances on and off may have been inefficient in this context. Alternatively, perhaps affordance-driven motor plans need to be maintained in order to prime subsequent actions, and such maintenance in turn depends on the goal-set of the individual. In either case, corresponding differences in attentional mechanisms may be critical. Future studies may provide further insights which of these models is most likely.