How are automatic processes elicited by intended actions?

Although unintentionality is one of the key elements common to most traditional definitions of automaticity (e.g., Posner and Snyder, 1975; Shiffrin and Schneider, 1977; Hasher and Zacks, 1979; Jacoby et al., 1993; Tzelgov, 1997; Moors and De Houwer, 2006), more recent approaches pointed out that many automatic processes are affected and even elicited by intended processes (e.g., Bargh, 1994; Hommel, 2000; Tzelgov and Ganor-Stern, 2005; Yamaguchi and Proctor, 2012). However, there was no attempt to provide a systematic analysis of the different ways in which intended processes activate automatic ones. The present paper aims to fill this gap. We view automatic processes as a product of interplay between information in long-term memory, reflecting life-long experience, and information in working memory, reflecting the requirement of the current situation. Such requirements are translated into task sets (Monsell et al., 2001), which activate the intended processes directly but may also activate some unintended processes indirectly. This activation of an unintended process (AUP) is a product of the overlap existing between the parameters of the intended process and those of unintended processes that are often highly practiced (Kornblum et al., 1990; Hommel, 2000). Automatic processes can be triggered by AUP, and in some cases they will not start without it. Automatic processes differ in their dependence on AUP. Some automatic processes need very little AUP to occur, while for others, stronger AUP is needed (e.g., Bargh, 1989, 1992). Furthermore, AUP may not only trigger the occurrence of a process, but may also enhance its behavioral manifestation. For example, when the interference paradigm is used, the behavioral manifestation of the automatic processing, which is the interference effect, will be larger in conditions of strong vs. weak AUP. We expand on the original taxonomy of dimensional overlap developed by Kornblum et al. (1990) and propose a classification of five levels of AUP, by which automatic processes are activated by intended ones. Furthermore, we demonstrate their effects using empirical findings of well-known indicators of automaticity. Note that a single task can provide AUP at multiple levels.