One key paradigm that has been used to investigate hypothesis-testing behaviour is Wason’s (1960) 2-4-6 task. This exists in two main forms: The standard version requires participants to induce a single rule, whilst the logically identical Dual Goal (DG) version, introduced by Tweney et al. (1980), asks participants to discover two related rules. In the standard version success rates of 20% are typically reported, whereas DG instructions increase solution rates to over 60%. One hypothesized explanation for this facilitated effect is the Goal Complementarity Theory (Wharton, Cheng & Wickens, 1993), which proposes that facilitated performance is mainly attributable to the complementary nature of the two unknown rules in the DG task. The present study investigated this theory by manipulating the to-bediscovered rules in order to produce both complementary and non-complementary DG versions of the task. Results did not lend support to the Goal Complementarity Theory. However, a close analysis of the triples generated by participants led to the formation of a new account of the facilitatory effect of DG instructions that centred on the role played by negative triples (and especially descending ones) in promoting task success.