Diagrama de Gantt (fragmento )

The aim of Experiment 1 was to determine the influence of pathway complexi- ty in terms of overall length as well as number and direction of consecutive turns along an outbound path on homing accuracy and response latency of Nonturners and Turners, using an allocentric or an egocentric reference frame, respectively. Since the primitive parameters of the reference frames differ, it was of particular interest whether these were updated in a history-free or configural manner, as measured on the basis of angular adjustments and response latencies.

Subjects were able to solve the task, as corroborated by the pronounced corre- lations between adjusted and expected arrow adjustment. Also, the absence of side effects proved egocentric and allocentric spatial representations to en- close right and left sides into a coherent representation space (Andersen, 1995; McNaughton et al., 1996). In several ways, the present experiment pro- vided strong evidence for the rejection of history-free updating during visual path integration. Number and direction of successive heading changes as well as laterality of end position revealed an impact on measures of homing accura-

cy and latency, suggesting configural updating (Fujita et al., 1993) of distance and direction during visual path integration. This was the case for both strate- gy groups using either an allocentric or an egocentric reference frame. Abso- lute error scores indicated that Nonturners and Turners were more accurate at inner end positions as compared to more lateral end positions. This deteriora- tion of accuracy was most pronounced for tunnels with two opposite turns ending at the outmost eccentricity, bearing resemblance to results of complex- ity-related increases of absolute error by Klatzky et al. (1990). With respect to the direction of error, specific patterns of under- and overestimations of eccen- tricity of end position were observed. Whereas inner end positions were over- estimated, more lateral end positions were underestimated. Interestingly, only when traversing tunnels with one turn the mean of adjusted homing arrows corresponded to the mean of the presented eccentricities, suggesting that this configuration was encoded more or less correct. For paths of higher complexi- ty, the mean of the adjusted values was shifted dependent on the relative di- rection of the second turn. Whereas eccentricities of tunnels with two turns bending to the same direction were overestimated, end positions of tunnels with opposite turns were underestimated. This linear shift only fits the encod- ing-error model of Fujita and colleagues (1993) under the assumption that en- coding of incoming visuo-spatial information is not a context-independent process but, rather, that it takes the previously traversed paths into considera- tion. Therefore, encoding is additionally determined by expectations of the ‘average’ pathway pervading the encoding process in a top-down manner, as proposed by Klatzky et al. (1999).

Both strategy groups displayed comparable error patterns although the under- lying primitive parameters should differ. For longer and more complex out- bound paths configural updating implies an increase of elements that have to be stored (and updated) as the navigator proceeds. Comparable error scores might therefore indicate that Turners and Nonturners encode the same amount of elements along the outbound trajectory, supporting the assumption of qualitative equivalence of representation systems (Avraamides, Loomis, Klatzky, & Golledge, 2004b; Burgess, 2006; Mou et al., 2006). Therefore, it seems reasonable to assume that within allocentric and egocentric reference frames identical objects are chosen for constructing an enduring representa- tion.

Response latencies were small overall, suggesting that updating of object bear- ings and distances was already accomplished during the passage (Riecke & Wiener, 2007). Importantly, Turners and Nonturners response latencies dif- fered. Nonturners’ response latency was found to be unaffected by pathway complexity, whereas Turners needed longer to initiate their homing response when paths contained more turns, particularly, when turns were bending into opposite directions. The complexity-related increase in response latency for Turners replicated results of Loomis et al. (1993). The update of multiple ego- centric distances and directions within an egocentric reference frame is com- putationally more demanding as compared to the update of allocentric direc-

Chapter 3 – Behavioral Analyses tions, since an increasing number of represented objects has to be updated with each successive step and rotation. Therefore, Turners might have taken longer to initiate the homing response since spatial relations within the ego- centric reference frame had to be updated until the very end of the passage, whereas for Nonturners the spatial layout might have been already present during the last segment (Riecke & Wiener, 2007).

Difficulties with opposite turns also became manifest in side error scores, rep- licating findings on side errors for more complex pathways of Riecke & Wiener (2007). The question was where these side errors emerge from. Post- experimental interviews revealed that subjects experienced increased difficul- ties when traversing tunnels with two opposite turns, resulting in high ambigu- ity and uncertainty with respect to the side the homing arrow should point to. This stands in marked contrast to real-world navigation, where subjects can distinguish right and left turns easily, even with eyes closed (Cornell & Greidanus, 2006). Since both strategy groups displayed increasing side errors, it seems plausible to assume that side errors arose at information-processing stages common to Turners and Nonturners, e.g., during the uptake and encod- ing of visuo-spatial information, as suggested by the configural updating mod- el of Fujita and colleagues (Fujita et al., 1993).