This framework is based on the target-array confusion model described by Dixon and distinguishes between the available perceptual information subjects have to make a response and the strategies subjects use to combine this information and arrive at a decision. It is assumed that variations in response strategy will affect primarily the relative number of misses and false alarms in this paradigm and will leave overall per cent correct largely unaffected. Instead, per cent correct will be primarily a function of the available perceptual information. RKSLIME Nous proposons un cadre de travail pour discuter Tcffet d'interference trouvc par Briand ct Klein (1988), Di Ix>llo et Moscovitch (1983), et Dixon (1986. Ce cadre de travail est base sur le modele de confusion cible-rangee d6crit par Dixon et fait la distinction entre ('information perceptuellc disponible a laquclle les sujets doivent rcpondre et les strategies que les sujets utilisent pour combiner ccttc information en vue d'arriver a une decision. Nous assumons que des variations dans la strategic de reponsc affecteront principalement le nombre d'erreurs ct dc fausses alarmes dans ce paradigme tout en laissant intact lc pourcentage global correct. Celuici sera principalement fonction de 1'information perceptuelle disponible.Recently, a number of researchers have reported an interference effect that occurs when subjects view a brief alphanumeric display followed by a single target item (Briand & Klein, 1988; Di Lollo & Moscovitch, 1983;Dixon, 1986; Dixon & Twillcy, 1988). Generally, it is found that subjects' ability to decide whether the target item was present in the array is poor when the target follows the array by 100-200 ms, but improves with either longer or shorter stimulus-onset asynchronies (SOAs). This pattern of performance is reminiscent of metacontrast masking, in that performance is a U-shaped function of SOA. However, as argued by Di Lollo and Moscovitch and by Dixon, the interference is unlikely to be due to the same kind of mechanism: The effect occurs over larger spatial and temporal intervals than has been reported for metacontrast and is unaffected by a variety of display parameters that are important in visual masking, such as the spacing of the array items or the distance between the array and the target. Instead, it has been proposed that the interference effect is due to some interaction between the array and the target at a more symbolic or conceptual level (Di Lollo & Moscovitch).Dixon (1986) reported an intriguing aspect of this interference effect: Generally, the errors that subjects make at short SOAs are false alarms, in which subjects claim *Preparation of this article was supported by Natural Sciences and Engineering Research Council of Canada Grant A8263. The author would like to thank h'red Morrison for comments on an earlier version. Address reprint requests to