The use of distinct reference frames as a means to represent entities in space leads to differences in the primitive parameters stored in the spatial represen- tation (Klatzky, 1998). The aim of the present investigation was to investigate whether homing accuracy dependent on the use of distinct reference frames is differentially influenced by path complexity. If reference frame specific para- meters are updated on a moment-to-moment basis, as proposed in history-free models of spatial updating(Benhamou & Seguinot, 1995; Hartmann & Wehner, 1995; Merkle et al., 2006; Müller & Wehner, 1988) overall bearings and dis- tances between the navigator’s current position and an arbitrary starting point should be available irrespective of path complexity (Loomis et al., 1999b). His- tory-free updating also implies that the navigator is neither able to retrace the path nor to determine the layout of the trajectory, as observed in arthropods (Merkle et al., 2006; Müller & Wehner, 1994, 2007). However, there is empiri- cal evidence that humans are capable of these tasks (Cornell & Greidanus, 2006; Cornell & Heth, 2004; Loomis et al., 1993). Therefore, it seems plausible to assume that the resulting representation contains more than just the dis- tance and direction to the starting point of an excursion.
The encoding-error model (Fujita et al., 1993) as representative of models based on configural updating proposes that spatial updating is not only ex- ecuted in constant intervals, but includes significant time-points of travel. Therefore, the resulting spatial representation contains not only information regarding overall distance and direction between current position and starting point, but also the layout of the traversed outbound path on the representa-
tional level. However, rotational information has been found to have a higher impact on updating as compared to translational information (Kearns et al., 2002; May, 2004). Therefore, heading changes constitute the critical time points of a passage where spatial updating is accomplished (Loomis et al., 1993; May & Klatzky, 2000; Riecke et al., 2002). Since turning locations are stored as points that determine the path layout, more complex pathways, i.e., containing more turns, require increasing cognitive resources due to the in- creasing number of relations to be encoded. This results in prolonged homing latencies and more pronounced errors during retrieval (Jansen-Osmann & Wiedenbauer, 2006; Kearns et al., 2002; Loomis et al., 1993; Maurer & Seguinot, 1995).
During history-free spatial updating the navigator has to refresh only overall return bearing and distance to the starting point. Both parameters are primi- tives within the egocentric and the allocentric locational representation. In contrast, configural updating within an egocentric reference frame requires the navigator to update egocentric return distance and bearing of the starting point, but also egocentric self-to-object relations (and distances) of all objects encountered along the outbound pathway. Since egocentric bearings vary with each directional change of the navigator’s intrinsic axis of orientation the ego- centric system is instable and updating of egocentric bearings becomes more difficult with increasing number of turns. More importantly, the direction of subsequent turns might impact egocentric updating, since represented ele- ments have to be shifted/rotated with respect to the intrinsic axis of the naviga- tor (see Figure 3.1).
Chapter 3 – Behavioral Analyses
Figure 3.1: (A) Configural updating within an egocentric reference frame. As the subject moves along the outbound trajectory, egocentric distances and bearing are updated at signifi- cant time points of travel, e.g., during rotations. Thus, the path is partitioned into straight segments separated by turns. During the initial translation from S to P1, only the return dis- tance has to be updated while the bearing remains identical. However, during the curve, bear- ing as well as distance change. As a result, S moves to the right side with respect to the navi- gator’s intrinsic axis of orientation. Between turns, distances and bearings have to be updated accordingly. During the curve at P2, the starting point again moves from the right to the left side. (B) By contrast, within an allocentric reference frame, traversing the outbound trajectory based on configural updating denotes the updating of allocentric distance and bearing with respect to the reference direction. In order to process rotational information, the navigator has to update the allocentric heading, i.e., the relative direction of travel with respect to the refer- ence direction. Since within an allocentric reference frame the navigator moves within the array of stationary objects, the return bearing does not change sides. If the allocentric bearing after the first turn is to the left, it is still on left as the path continues with an opposing turn. Similarly, increasing errors due to increasing number of elements should also be present for allocentric configural coding. However, since within the allocen- tric reference frame the navigator’s changes in orientation are computed with
respect to an external reference direction, providing information about the al- locentric heading of the moving agent, the relative direction of changes in the navigator’s intrinsic axis should have no impact on allocentric coding. Thus, within an allocentric reference frame successive heading changes should have no impact on accuracy and response latencies.
Taken together, whereas history-free updating supposedly results in compara- ble errors and response latencies for homing responses based on distinct ref- erence frames, configural updating should lead to strategy-specific deteriora- tions due to differences in updating of primitive parameters.