In addition to there being methodological concerns as to the measurement of automatically activated implicit knowledge, inconsistent findings across studies have also prompted an assessment of prevailing stereotype theory (Bargh, 1994; 1998; von Hippel, Sekaquaptewa, & Vargas, 1995). It was noted by Bargh (1994) that research findings suggested more variation in stereotypic processing than Devine's automatic versus controlled conceptualisation allowed (von Hippel et al., 1995). This prompted the development of an alternative thesis. From the outset Bargh argued that the inconsistent findings may be attributed to the fact that different procedures serve to tap differing cognitive components involved in the automatic activation of stereotype-related knowledge (Bargh, 1994). His reconceptualisation essentially holds that rather than occurring in an all-or-none fashion, there are sub-components of automatic and controlled processing that combine to produce particular responses in the presence of certain conditions (Bargh, 1994; 1998). Bargh proposes that three principal components, awareness, intentionality, and controllability are identifiable in the course of knowledge activation and that each of these features has an automatic state and a controlled state.
In Bargh's thesis the first component, awareness, refers to the extent to which thought and behaviour occur outside awareness. When an individual is unaware of the target or associations with the target, the mental process is considered to be automatic. Intentionality concerns whether an individual is in control of the instigation of the mental process. In an automatic process, the 'start-up' or initiation of the process is unintentional. The controllability element refers to the individual's ability to override or control the influence of the process if so desired. Clearly, an automatic process would be one that an individual is unable to terminate once commenced. Hence, in contrast to Devine's model, Bargh's thesis supports a more multi-component view of the processing of information.
With respect to the activation of implicit stereotypic information, Bargh argues that once relevant group features are encountered, the activation of related information occurs preconsciously, that is, the processing of information is triggered automatically in a manner which is unintentional, uncontrolled and is outside of awareness (Bargh, 1994; Bargh, Chen & Burrows, 1996).
In order to explore the ways in which certain measurement paradigms can be seen to be operationalising elements of Bargh's information processing model, an illustration of Bargh's model (Table 3) will be explored with due reference to the previously discussed priming paradigms.
|No||Null set*||Null set||Null set||D|
Note. Null sets represent combinations that in Bargh's sense would not be possible.
The set labelled A (Table 3) refers to controlled processing, such as that which is measured in attitudinal scales and free response measures of cultural knowledge. Sets B and C refer to situations in which there may be awareness, but there may be a difference in the degree to which the process can be controlled once initiated (e.g., Augostinous et al.'s response latency task, 1994; Locke et al.'s Stroop task with longer stimulus-onset asynchrony). With respect to D, as there is no conscious awareness, it can been argued that the process is also unintentional and uncontrollable (e.g., the parafoveal priming paradigm employed by Devine 1989, and the lexical decision tasks employed by Banaji & Hardin, 1996). While this discussion may elucidate the way in which Bargh's criteria for the automaticity of cognitive processing can be operationalised, Bargh actually fails to stipulate whether all three or a combination of these features needs to be present in order for implicit knowledge to be activated.