Attention and Relations in Conceptual Combination

Ramm, Brentyn (2006). Attention and Relations in Conceptual Combination PhD Thesis, School of Psychology, University of Queensland.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
n01front_ramm.pdf n01front_ramm.pdf application/pdf 290.13KB 0
n02content_ramm.pdf n02content_ramm.pdf application/pdf 1.96MB 1
Author Ramm, Brentyn
Thesis Title Attention and Relations in Conceptual Combination
School, Centre or Institute School of Psychology
Institution University of Queensland
Publication date 2006
Thesis type PhD Thesis
Supervisor Professor Graeme Halford
Abstract/Summary Combined concepts such as computer chair, road rage, and terrorist threat are common in everyday language. Combining concepts into novel meanings is achieved spontaneously in conversation, and serves to decrease working memory loads. Due to this function, it could be argued that conceptual combination is a process similar to chunking. Yet despite the importance of conceptual combination to human cognition, and the assumption that chunking occurs in working memory (e.g., A. D. Baddeley & G. Hitch, 1974), no previous studies have assessed the role of attention in the process of conceptual combination. The Competition Among Relations in Nominals (CARIN) theory (C. L. Gagne & E. J. Shoben, 1997) proposes that relations provide the mortar for combining concepts. Furthermore, these relations compete to become the binding link between concepts, with priority given to relations that historically have been frequently linked with the modifier (Relational Frequency). Since evidence has shown that the binding of relations is a capacity-constrained process (e.g., G. S. Halford, W. H. Wilson & S. Philips, 1998; N. Robin & K. J. Holyoak, 1995) this thesis seeks to elucidate the role of attention and relations in conceptual combination. The attentional demands of combining novel modifier-noun phrases are explored in Experiment 1 using a dual-task methodology (digit transformation). The results show that a concurrent task significantly increases the response time of a sense-nonsense decision, supporting the hypothesis that conceptual combination is capacity-limited. However, it was not clear from the results whether it is the simultaneous activation of the word meanings, the combination process itself, or both that requires attention. To assess this issue, Experiment 2 used probe reaction time (RT) as an indication of the attentional demands of a sense-nonsense decision or a lexical decision to the word-pairs. The results show that the process of conceptual combination is attentionally demanding after accounting for the demands of activating the constituent meanings of the wordpairs. No effect of Relational Frequency was found on attention in this experiment. However, each participant only made a sense-nonsense decision to half of the phrases, therefore the variability of scores was poor for assessing this issue. Experiment 3, therefore, again assessed probe RT but used a sense-nonsense decision for all of the word-pairs. Results show that the linking of concepts with a low frequency relation is more attentionally demanding than linking with a relation of high frequency. Further investigations used Google hit rates as an indication of word-pair familiarity, and show contrary to E. J Wisniewski and G. L Murphy (2005) that familiarity of the word-pair does not account for Relational Frequency. A new metric of relations (Relational Familiarity) was developed based upon the median Google score of each relational type. While both Relational Frequency and Relational Familiarity provided equal contributions to the accuracy of the sense-nonsense decisions, frequency of the relational type was subsumed by familiarity in predicting probe RT. That is, Relational Familiarity provided the best predictor of the capacity-demands of conceptual combination, thus showing that it is easier to select relations of high familiarity than low familiarity. Inhibition was suggested as a viable candidate for this selective mechanism. Lastly, relation competition was shown to be a poor predictor of probe RT scores. These findings, in conjunction with previous research, support a Capacity-Constrained Relation Selection model of conceptual combination, including four characteristics: 1. Relations compete in parallel in the activated proportion of long-term memory to enter the focus of attention. 2. The activation and maintenance of the meanings of the individual word-pairs is attentionally demanding. 3. The selection of relations is also a capacity-constrained process. 4. The difficulty of selecting a relation depends upon its familiarity rather than its frequency. This project extended the CARIN model (C. L. Gagne & E. J. Shoben, 1997) to show that attention has a function in both forming new representations and in the selection of relations. Overall, it was demonstrated that attention plays an essential role in forming combined concepts and thus should be included in any valid model of this process.

Citation counts: Google Scholar Search Google Scholar
Access Statistics: 136 Abstract Views, 1 File Downloads  -  Detailed Statistics
Created: Fri, 21 Nov 2008, 15:36:48 EST