Tipultech logo

Measuring Implicit Knowledge: Assumptions So Far

Sarah Ogilvie

At this stage researchers investigating stereotyping and the underpinning cognitive processes have failed to reach a consensus concerning accurate measures of implicit knowledge (Bargh, 1994; Wittenbrink et al., 1997), although the different measures are based on a set of widely shared assumptions. The principal assumption, one that is widely accepted in the fields of both cognitive and social psychology, is that the speed and depth of processing information is influenced by previously encountered information (Hilton & von Hippel, 1996). Hence exposure to particular information will arguably serve to prime or activate related information, which is stored in memory. While priming effects can be achieved in many ways there are generally two main methods that are employed.

The first method employed aims to temporarily increase the accessibility of available knowledge (Higgins, 1998). This is typically accomplished by frequently priming the stored knowledge with related priming material. It has been found that increasing the frequency with which information is primed, increases the temporary accessibility of stored information (Higgins, 1998). Methods such as Devine's (1989), in which the frequency of priming is increased across conditions, allow a comparison of the effects of frequent priming. When differential effects are obtained across conditions it is assumed that the priming material has increased the accessibility of related stored information (Higgins, 1998).

The second way in which priming techniques are typically employed is to access chronically available knowledge by recent priming (Higgins, 1998). In this it is assumed that if certain knowledge is chronically available then pairing the priming material with a related stimulus as opposed to an unrelated stimulus should produce differing responses (Higgins, 1998). Experimental procedures conventionally employed to assess the presence of knowledge related to the priming material, involve priming with a stereotype or category label, after which a stereotype-related or -unrelated trait is presented (e.g., Blair & Banaji, 1996; Dovidio et al., 1986; Dijksterhuis & van Knippenberg, 1996; Foddy, 1997; Macrae, Milne & Bodenhausen, 1994; Stapel & Koomen, 1998; Stangor & Lange, 1994; Stroupe & White, 1996; Wittenbrink et al., 1997). If the stereotype or category prime has served to activate related knowledge, subsequent exposure to information related to this knowledge (a stereotype-related trait) should elicit a response that differs to exposure to unrelated information (a stereotype-unrelated trait).

In stereotype research it is further assumed that by regulating the length of time a prime is exposed, and the level to which the prime is masked, activation of attentional, or controlled cognitive processes can also be regulated (Hilton & von Hippel, 1996; Klinger & Greenwald, 1995; Locke et al., 1994). Controlling the time from onset of the prime to onset of the stimulus, or the stimulus-onset asynchrony (SOA), allows control over whether the respondent is able to detect the prime. Masking procedures are often used to control both the detection and whether or not the prime is recognised as a particular word (Klinger & Greenwald, 1995; Reber, 1995). A type of masking often employed involves presentation of the target word directly after the prime (Stroupe & White, 1996). In this procedure the target word acts as backward mask on the prime, and dependent on the SOA, it can also serve to reduce the residual effects of repeated exposure to a particular prime (Forster & Davis, 1984; Reber, 1995).

In activating chronically accessible implicit knowledge, results have also been found to vary according to the type of decision that follows the priming manipulation (Forster & Davis, 1984; Klinger & Greenwald, 1995; Stroupe & White, 1996). In order to ascertain whether the priming has facilitated activation of the semantic content of the target, researchers incorporate decision tasks that require a judgement to be made after exposure to each prime-target pair. In employing tasks that require a decision which is unrelated to the relationship between the prime-target pair (e.g., Banaji & Hardin's lexical decision tasks, 1996), evidence suggests that implicit knowledge is less accessible than when the task is related to the prime-target relationship.

Tasks that incorporate decisions which question the prime-target relationship commonly require participants to indicate whether a semantic association exists between primes and targets (Blair & Banaji, 1996; Dovidio et al., 1986; Klinger & Greenwald, 1995). It has been found that by including such a task, priming procedures that elicit unrecognisable primes fail to facilitate such a decision (Dovidio et al., 1986; G. Yelland, personal communication, October 13, 1998). Accordingly, it is argued that when a semantic decision is called for, a reduction in the level of masking or an increase in the stimulus-onset asynchrony, is needed to allow for recognition of the prime and subsequent determination of its relationship to the target (G. Yelland, personal communication, October 13, 1998).

<< >>