Forward and backward blocking of causal judgment is enhanced by additivity of effect magnitude

Abstract

When two causes for a given effect are simultaneously presented, it is natural to expect an effect of greater magnitude. However many laboratory tasks preclude such an additivity rule by imposing a ceiling on effect magnitude-for example, by using a binary outcome. Under these conditions, a compound of two causal cues cannot be distinguished from a compound of one causal cue and one noncausal cue. Two experiments tested the effect of additivity on cue competition. Significant but weak forward blocking and no backward blocking were observed in a conventional "allergy" causal judgment task. Explicit pretraining of magnitude additivity produced strong and significant forward and backward blocking. Additivity pretraining was found to be unnecessary for another cue competition effect, release from overshadowing, which does not logically depend on additivity. The results confirm that blocking is constrained when effect magnitude is constrained and provide support for an inferential account of cue competition.

Additivity pretraining and cue competition effects: Developmental evidence for a reasoning-based account of causal learning.

“…This finding is consistent with those of other recent studies with adults that suggest that blocking is either absent or greatly reduced when pretraining is nonadditive (Beckers et al, 2(X)5;Lovibond et al, 2003). Nevertheless, we recognize that it is possible for blocking to occur after nonadditive pretraining conditions, particularly under forward presentation conditions, (Beckers et al, 2005;Lovibond et al, 2003), and indeed backward and forward blocking has previously been reported after nonadditive pretraining conditions in a study of 5-year-olds (McCormack et al, 2009). This raises the question of whether a reasoning account is appropriate to explain all instances of this cue competition effect.…”

“…Demonstrations of blocking (or related phenomena) have been provided in various species (potentially including invertebrates such as snails, Acebes, Solar, Camero, & Loy, 2009, and honey bees, although see Blaser, Couvillon, & Bitterman, 2(X)6), and animal researchers would naturally be reluctant to assume that all of these species are capable of inferential reasoning. Beckers et al (2005) and Lovibond et al (2003) allow for the possibility that there may not be a single set of processes underpinning blocking, and that, even in humans, lower-level associative processes may be involved under some circumstances. However, taken together, the findings of the current study strongly suggest that explaining the effects of additivity pretraining on blocking does require invoking a role for inferential reasoning.…”

“…Engaging in this chain of reasoning will disambiguate B's causal status, giving low causal ratings for B and thus producing a blocking effect. Such reasoning is not available when causal outcomes are not additive; thus, blocking effects after nonadditive training are likely to be much weaker or nonsignificant (Beckers et al, 2005(Beckers et al, , 2006Lovibond et al, 2003).…”

Inferential dependencies in causal inference: A comparison of belief-distribution and associative approaches.

“…It may be surprising that our experiment found evidence for backward blocking (and its preventive variant), given that recovery from overshadowing is often observed in situations where backward blocking is not (Coriett et al, 2004;Larkin et al, 1998; see also Beckers, De Houwer, Pineno, & Miller, 2005;Lovibond, Been, Mitchell, Bouton, & Frohardt, 2003;. How are we to reconcile our finding (and similar findings such as Wasserman & Berglan, 1998;Wasserman & Castro, 2005) with these other findings?…”

Associative repetition priming as a measure of human contingency learning: Evidence of forward and backward blocking.

“…In fact, one of the main strengths of this theory is based on the empirical evidence consistent with predictions regarding the circumstances in which blocking should be found and the circumstances in which blocking should be rather unlikely. Specifically, the blocking inference should be rather unlikely if (a) the relationships between cues and outcomes are arbitrary rather than causal (De Houwer, Beckers, & Glautier, 2002;Waldmann, 2000), (b) the outcomes are always present at their maximal level, and the additivity assumption cannot be applied (Lovibond, Been, Mitchell, Bouton, & Frohardt, 2003;Mitchell & Lovibond, 2002;Mitchell et al, 2005), and (c) the goal that has to be achieved at test does not encourage participants to apply the blocking inference (e.g., Beckers et al, 2005;De Houwer et al, 2007;see also De Houwer, 2009, for a review). Because our task meets all of these conditions, the observation of blocking due to the application of the blocking inference should be very improbable.…”