Arithmetic priming from neglected numbers

Abstract

Patient AM, with left visuospatial neglect, and 31 healthy participants performed a parity judgment task on numbers presented to their right parafovea. Target numbers were preceded by a pair of digits (prime) appearing peripherally for 100 ms either in their left (LVF) or in their right visual field (RVF), which participants had to ignore. In healthy participants, when primes were arithmetically related to the following target, performance was significantly slower than when primes and targets were not related. In contrast, AM's performance was slower in the related than in the control condition when prime digits appeared in his RVF, whereas it was faster in the related than in the control condition when prime digits appeared in his LVF. This suggests that neglected numbers were nevertheless processed at least until the level of stored arithmetic knowledge. Thus, visuospatial neglect does not prevent neglected numbers from accessing their representations in arithmetic networks, which seems to be a highly automatised skill. Moreover, AM's pattern of data (i.e., interference from RVF primes vs. facilitation from LVF primes) supports the hypothesis of a link between conscious attention and inhibitory processes, as proposed by Fuentes and Humphreys (1996).

Acknowledgments

Preparation of this manuscript was supported in part by grants from the European Commission (RTN grant HPRN-CT-2000-00076) to CU and from the University of Padova (Progetto d'Ateneo CPDA034753) to MLR. We would like to thank Chiara Fiaccadori for her help in data collection, and patient AM and our healthy participants for their collaboration. We are grateful to the doctors of the Department of Neuroscience of Vicenza for permission to investigate their patients, and to Prof. Vittore Pinna, Head of the Department of Neuroradiology of Vicenza, for having kindly provided AM's CT scan.

Notes

Methods for testing whether the difference between two experimental conditions deviates significantly from that found with an ad hoc control sample either treat the control sample statistics as parameters (i.e., they assume the control sample to be a population) or use modified t-tests. Monte Carlo simulations demonstrated that both these methods lead to undesirable increments of the Type I error rate (Crawford & Garthwaite, in press). In contrast, Crawford and Garthwaite's Revised Standardized Difference Test treats control sample statistics as sample statistics and achieves good control of the error rate. Furthermore, a specific program (DISSOCS.exe) has been developed to test for the presence of either classical or strong dissociation when scores are available on two different experimental conditions (e.g., X and Y) both for the patient and for the control sample. The definition of classical dissociation requires that a patient show a deficit on X but be within normal limits on Y (Shallice, 1988). Crawford, Garthwaite, and Gray (2003) proposed formal criteria for a classical dissociation that, in addition to the standard requirement of a deficit on X and normal performance on Y, incorporated the requirement that patient's performance on X be significantly poorer than performance on Y, which provides a positive test for dissociation (thereby lessening reliance on proving a null hypothesis – i.e., the hypothesis of no difference between patient and control sample as for performance on Y). Thus, we compared AM's priming effects in LVF and RVF with those of healthy participants through Crawford and Garthwaite's (in press) procedure.

The test was performed with the program DISSOCS.exe which can be downloaded freely from the following address: http://www.abdn.ac.uk/∼psy086/dept/SingleCaseMethodsComputerPrograms.htm