The present study investigates the joint role of spatial ability, imagery strategy and visuospatial working memory (VSWM) in spatial text processing. A set of 180 participants, half of them trained on the use of imagery strategy (training vs no-training groups), was further divided according to participants' high or low mental rotation ability (HMR vs LMR). Each group listened to environment descriptions and performed recall tasks before and immediately after training/no-training, and again after listening to a text while performing a spatial tapping task. Visuospatial and verbal tests were also administered. Our results showed that HMR had a better spatial profile than LMR participants, and that only LMR participants benefited from training and showed the interference effect. Overall, our findings indicate that a good spatial ability reduces the spatial interference effect and that poor spatial ability, which is related to the spatial interference effect, can be partially compensated by learning imagery strategy.
In everyday life, one way to convey environmental information is to use descriptions. A person looking for a given destination may, for instance, identify it by reading a tourist guide book illustrating the path to take or by asking someone on the street for spatial directions. The readers or listeners usually reach the right destination if they construct a mental representation of the environment described. The result is a mental model (Johnson-Laird, 1983 and van Dijk and Kintsch, 1983) in which spatial content is represented, preserving the physical properties of space, such as the relative relationships between objects (Perrig and Kintsch, 1985 and Taylor and Tversky, 1992) and information about distances (Rinck, Hahnel, Bower, & Glowalla, 1997).
In the last decade, a number of studies have focused on analyzing the cognitive functions, such as working memory (WM), involved in the construction of mental representations from spatial descriptions (of an environment). WM involvement in processing spatial descriptions has also been investigated in relation to spatial ability or imagery strategy, but how these aspects work together remained unexplored. The novel goal of the present study was to shed light on how spatial ability and strategy interact in the processing of spatial descriptions, analyzed in relation to WM.
3.1. Selection of HMR and LMR
Two groups were selected on the basis of their performance in the MRT. Given the gender differences in MRT performance (Linn & Petersen, 1985), as confirmed in our sample—F (1, 178) = 33,62 η2 = .16, p ≤ .001—where males (M = 10.60, SD = 4.29) outperformed females (M = 7.07, SD = 3.85), the selection of high and low MR individuals was done separately for males and females (as in Meneghetti et al., 2009) using percentile criteria (25th and 75th percentiles). Females with a MRT score ≤ 4 (corresponding to the 25th percentile) joined the LMR group, and those with a MRT ≥ 9 (corresponding to the 75th percentile) joined the HMR group; males with a MRT ≤ 8 (25th percentile) were assigned to the LMR group and those with a MRT ≥ 13 (75th percentile) to the HMR group. Overall, 57 HMR and 57 LMR participants (with 26 males and 31 females in each group) were selected, who differed in their MR scores (F (1, 112) = 328.52 η2 = .75, p ≤ .001; HMR: M = 13.35 SD = 3.09; LMR: M = 4.25 SD = 2.21). The participants in the training and no-training groups were divided as follows: 29 trained and 28 untrained HMR individuals, and 29 trained and 28 untrained LMR individuals.
3.2. Visuospatial and verbal measures
To analyze the profiles of the HMR and LMR individuals, the two groups were compared in visuospatial and (for control purposes) in verbal measures. Multivariate analysis showed a general significant effect, F (12, 101) = 328,52 η2 = .35, p ≤ .001; in particular, univariate ANOVA indicated that HMR outperformed LMR individuals in the following measures (adopting Bonferroni's correction, only p ≤ .01 was considered significant): Knowledge and use of cardinal points (in SDSR), F (1, 112) = 6,33 η2 = .05, p ≤ .01; Survey representation (in SDSR), F (1, 112) = 13,52 η2 = .11, p ≤ .001; Spatial imagery preference (in OSIQ), F (1, 112) = 7,10 η2 = .06, p ≤ .01; and backward versions of the Corsi blocks task, F (1, 112) = 17,54 η2 = .14, p ≤ .001, and Digit span task, F (1, 112) = 9,33 η2 = .08, p ≤ .01. No significant differences were found for the other measures (F < 1 to F = 3.20 p = .08) (Table 1.).
