A Modified SpAM Task for Investigating Typicality and Frequency Effects during Category Processing
Loading...
Author(s)
Koenen, Reba
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
This dissertation used three tasks to investigate categorization measures and the strength of category norms. Category norms are rank orderings of the most common exemplars (i.e., apple) which belong to a given category (i.e., Fruits). Over the past several decades, both cognitive psychologists and cognitive scientists have studied how people engage in category processing. The most common method used is the production task and its measure of frequency – the number of times an exemplar is produced across participants for a given category (e.g., Battig & Montague, 1969). An alternate, classic method is the rating task and its measure of typicality – the rating of how well an exemplar represents its category (e.g., Rosch, 1975). The third task used to inform studies of categorization is the spatial arrangement method (SpAM) and its similarity measure of position – the location of an exemplar, relative to other exemplars of that category, on a 2D plane (e.g., Goldstone, 1994a). In the current study, participants completed these three tasks (i.e., production, rating, a modified version of SpAM) to address three overarching research goals.
The first goal was to investigate the relationship between the frequency and typicality effects. I asked whether both measures are comparable indices of the centrality of exemplars to categories, as the concepts and categorization literature suggests (e.g., Banks and Connell, 2022). I predicted that these measures, and their tasks (i.e., production, rating), are more distinct than previously assumed. There was mixed evidence for this hypothesis. Correlations between the frequency and typicality data from both the current and prior studies resulted in larger correlations (r .40) than predicted.
The second goal was to examine the usefulness of the modified SpAM task, its standard measure of position, and the novel measure introduced here of placement-order (i.e., the order in which participants place exemplars onto the 2D plane). I asked whether either of these measures, and the modified SPaM task more generally, could be used as an alternative method to the production task and its frequency data or the typicality task and its rating data. The modified SpAM task achieves the primary advantage of each of these classic tasks: it is short in duration, like the production task, and it provides information about a complete list of exemplars for a category, like the rating task. I predicted that correlations between both the position and placement-order data, with the frequency and typicality data from the current and prior studies, would be moderate or large (r .30). There was weak evidence for this hypothesis. There were more small correlations (r < .30) than expected, especially with the placement-order data.
The third goal was to test the novel hypothesis that the frequency and typicality effects emerge at different timepoints during category processing. To do so, I used the production-order variable (i.e., the order in which exemplars are produced during the production task). This measure was adopted from the category fluency task which is used to investigate the claim that more frequent exemplars are more accessible in semantic memory, and, thus, produced earlier in time (Hills et al., 2012). I predicted that in the modified SpAM task, the most frequent exemplars would be placed earlier in time and that more typical exemplars would be placed slightly later. There was mixed evidence for this hypothesis. Placement-order timing intervals correlated best with the frequency, and then the typicality data from the current study. This was not the case with prior data, where the correlations with placement-order timing and both frequency and typicality tended to be small (r < .30). This finding could have been a result of the within-subjects design of this study. All participants (N = 90) completed each of the tasks three times each, with the three experimental categories used in the current study (i.e., Birds, Fruits, Vegetables).
Where the evidence did not support the hypotheses, there are several possible explanations. This gap may have been a consequence of the study’s novel design (completing all three categorization tasks within-subjects has not been done in prior studies). It may also have been a result of the small number of experimental categories (i.e., three) or participants’ unfamiliarity with some exemplars from the Birds category. Though these design decisions had their limitations, they also allowed a direct comparison of the tasks (i.e., production, rating, modified SpAM) and their measures (i.e., frequency, typicality, position). There was a significant difference in participants’ ratings of how well the different tasks captured their category structures, with participants ranking the rating task as more representative of their category structure than the production task. The modified SpAM task also had higher category structure representation ratings than the production task, though this difference did not reach statistical significance. These findings should guide future studies of categorization as they suggest there are limitations to the production task despite its efficiency (e.g., Banks & Connell, 2022).
Sponsor
Date
2024-12-05
Extent
Resource Type
Text
Resource Subtype
Dissertation