Most theories of analogical transfer focus on similarities between the learning and transfer domains, where transfer is more likely between domains that share common surface features, similar elements, or common interpretations of structure. We suggest that characteristics of the learning instantiation alone can give rise to different levels of transfer. We propose that concreteness of the learning instantiation can hinder analogical transfer of well-defined structured concepts, such as mathematical concepts. We operationalize the term concreteness as the amount of information communicated through a specific instantiation of a concept. The 5 reported experiments with undergraduate students tested the hypothesis by presenting participants with the concept of a commutative mathematical group of order 3. The experiments varied the level of concreteness of the training instantiation and measured transfer of learning to a new instantiation. The results support the hypothesis, demonstrating better transfer from more generic instantiations (i.e., ones that communicate minimal extraneous information) than from more concrete instantiations. Specifically, concreteness was found to create an obstacle to successful structural alignment across domains, whereas generic instantiations led to spontaneous structural alignment. These findings have important implications for the theory of learning and transfer and practical implications for the design of educational material. Although some concreteness may activate prior knowledge and perhaps offer a leg up in the learning process, this benefit may come at the cost of transfer.
Flexible Attention to Labels and Appearances in Early Induction Anna V. Fisher (fisher49@andrew.cmu.edu) Department of Psychology, Carnegie Mellon University 5000 Forbes Ave, Pittsburgh, 15213 USA Vladimir M. Sloutsky (sloutsky.1@osu.edu) Center for Cognitive Science, The Ohio State University 1961 Tuttle Park Place, Columbus, OH, 43210 1999; Sloutsky & Fisher, 2004), with induction being a function of the overall similarity of presented entities. In short, whereas both positions predict the importance of labels, they posit vastly different mechanisms. According to the former, the importance of labels stems from top-down conceptual influences, whereas according to the latter it stems from a low-level attentional mechanism. One way of distinguishing between the two positions is to examine the flexibility of reliance on different sources of information. If labels are more theoretically central than appearances, then it should be difficult if not impossible to change the reliance on labels in the course of associative learning. Conversely, if reliance on labels and appearances stems from attention being automatically allocated to these predictors, then changing attentional weights of labels or appearances should affect the reliance on these predictors in a subsequent induction task. Preliminary evidence that early in development attentional weights of various attributes are flexible comes from the research by Smith, Jones, and Landau (1996). In this study 2- and 3- year-old children were presented with a task in which they had to generalize a novel label to other objects. Children generalized labels based on shape and texture attributes if presented objects had eyes (presence of eyes is highly correlated with animacy), whereas participants relied solely on shape if the same objects were presented without eyes. Similarly, young children generalize names of solid substances along the lines of shape similarity, whereas names for non-solid substances were generalized along the lines of texture similarity (Samuelson & Smith, 1999). Furthermore, there is evidence that these attentional biases are shaped by experience, since they affect learning only after children have acquired many names for solid and non-solid substances (Samuelson & Smith, 1999). One way of changing an attentional weight of a predictor is by manipulating its predictive value. There is ample body of evidence in the animal learning literature indicating that if a cue is non-predictive in an associative learning task, reliance on this cue attenuates in a subsequent task (see Hall, 1991, for a review). Sloutsky & Spino (2004) demonstrated that the same is the case with 5-year- olds, with effects of associative learning sustaining for over two months. In sum, when performing induction, young children rely on both, appearance and label information. Some have argued that reliance on labels stems from top-down Abstract Young children have been demonstrated to rely on both labeling and appearance information when performing induction. According to some accounts, labels are more conceptually important than appearances. According to others, reliance on labels and appearances stems from a low- level attentional mechanism. The latter, but not the former, predicts flexible attentional shifts in reliance on labels or appearances. Results of the two reported experiments indicate that attention to labels and appearances can be flexibly modified through associative training, thus supporting the latter, but not the former account. Introduction The ability to perform inductive generalizations is crucial for acquiring new knowledge. For instance, upon learning that a particular poodle uses enzymes to digest food, one could generalize this knowledge to other poodles, other canines, and possibly other mammals. It is well-documented that the ability to perform simple generalizations develops early in life (Gelman & Markman, 1986; Sloutsky & Fisher, 2004a, 2004b; Welder & Graham, 2001), however specific mechanisms underlying early induction remain unclear. It has been demonstrated that when performing induction, young children rely on various sources of information, such as perceptual similarity and linguistic labels (Gelman & Markman, 1986; Sloutsky, Lo, & Fisher, 2001). Some have argued that even young children assume that these sources form a conceptual hierarchy, with some object properties being more important, or central, than others for determining category membership and generalizing properties of natural kind objects (Keil, et al, 1998; Gelman & Coley, 1991). For example, category labels are said to be proxies of category essences (e.g., Gelman & Coley, 1991), and as such their contribution to induction should be greater than that of peripheral properties, such as appearances. Others have argued that reliance on labels during induction may stem from a low-level attentional mechanism. In particular, under many conditions auditory input (including labels) overshadows (or attenuates processing of) corresponding visual input (Sloutsky & Napolitano, 2003; Napolitano & Sloutsky, 2004; Robinson & Sloutsky, 2004). As a result of overshadowing, entities that share the same label are perceived as more similar than the same entities presented without a label (Sloutsky & Lo,
The ability to perform induction appears early; however, underlying mechanisms remain unclear. Some argue that early induction is category based, whereas others suggest that early induction is similarity based. Category‐ and similarity‐based induction should result in different memory traces and thus in different memory accuracy. Performing induction resulted in low memory accuracy in adults and 11‐year‐olds, whereas 5‐, and 7‐year‐olds were highly accurate (Experiment 1). After training to perform category‐based induction, 5‐ and 7‐year‐olds exhibited patterns of accuracy similar to those of adults (Experiment 2). Furthermore, 7‐year‐olds, but not 5‐year‐olds, retained this training over time (Experiment 3). With novel categories, even adults performed similarity‐based induction, exhibiting high memory accuracy (Experiment 4). These results suggest a gradual transition from similarity‐ to category‐based induction with familiar categories.
In visual perception, the brain faces a difficult dimensionality reduction problem, with many nerve fibers converging to relatively few perceptual features. This is possible because meaningful structure in visual input lies on a lower-dimensional psychological manifold extracted by the visual system. However, while many cognitive processes are hypothesized to use geometric relationships between stimuli, there is no general method to assign psychological manifolds a geometry with an independently-verified notion of distance. To this end, we propose a model of the geometry of psychological manifolds that combines perceptual models with an attentional process that modifies discriminability according to context. These constraints induce a metric tensor that measures the length of curves on the manifold, including those corresponding to stimuli that change continuously through time. To investigate whether curve length corresponds to perceptual distance, we conducted an experiment (N = 39) comparing model predictions to judgements of the rate of change for dynamic stimuli with a one second duration. We first showed participants static, Gabor-like stimuli varying in spatial frequency and orientation, instructing them to place stimuli on a line according to the value of each dimension. We then presented pairs of changing stimuli and participants reported their confidence in which was changing faster in frequency or orientation with a slider. Using responses to static stimuli and assuming 80% attention to the cued dimension, difference in curve length predicted judgements for 36 of 39 participants (logistic GLM, >95% posterior probability). We then fit our model to all data and found it predicted judgements better than a flat approximation with one fewer parameter for 18 participants (approximate leave-one-out cross validation). These results suggest that distance on a psychological manifold is a meaningful measure of subjective change. Additionally, our second analysis suggests that flat approximations of psychological manifolds are inappropriate for some participants.
Attention and cross-modal processing: Evidence from heart rate analyses Christopher W. Robinson (robinson.777@osu.edu) Center for Cognitive Science The Ohio State University 208F Ohio Stadium East, 1961 Tuttle Park Place Columbus, OH 43210, USA Vladimir M. Sloutsky (sloutsky.1@osu.edu) Center for Cognitive Science The Ohio State University 208C Ohio Stadium East, 1961 Tuttle Park Place Columbus, OH 43210, USA dominance effect” (Lewkowicz, 1988a; 1988b; Robinson & Sloutsky, 2004; 2007; 2010; Sloutsky & Napolitano, 2003; Sloutsky & Robinson, 2008). Although the asymmetry between infant and adult literatures may reflect genuine developmental differences, it is also possible that the asymmetry stems from a lack of appropriate measure of auditory processing. In particular, most infant studies use visual fixations to examine auditory and cross-modal processing. For example, infants in many of the studies reported above were familiarized or habituated to an auditory stimulus, visual stimulus, or to a cross-modal stimulus. Infants in the cross-modal condition often failed to increase looking to a novel visual stimulus when it was paired with an old sound, suggesting that they did not discriminate the visual stimuli. This finding is noteworthy given that infants ably discriminated the same visual stimuli when they were presented unimodally. In contrast, there were no costs of cross-modal presentation on auditory processing: infants equally discriminated auditory stimuli when presented unimodally and cross-modally. However, auditory processing was never measured independently of visual processing (i.e., auditory processing was assessed by examining infants’ visual fixations). In the absence of a true measure of auditory processing, it is possible that visual dominance was missed, with visual input interfering more with auditory input than the reverse. The goal of the present research was to address this issue. The achievement of this goal requires an amodal measure of auditory processing. While sucking procedures and ERP tasks can provide modality- independent measures of auditory processing (e.g., Eimas, Siqueland, Jusczyk, & Vigorito, 1971; Nelson & deRegnier, 1992), sucking procedures are not appropriate for older infants and children and ERP tasks often require a large amount of trials. The present study uses infants’ Heart Rate (HR) to Abstract The study of cross-modal processing has generated two seemingly contradictory sets of findings. Studies examining cross-modal processing in infants often find evidence that auditory input interferes with visual processing, whereas studies with adults often find evidence for visual input interfering with auditory processing. However, in the absence of amodal measures of auditory processing, it is possible that visual input also interferes with auditory processing in young infants. The primary goal of the current study was to examine this issue by focusing on Heart Rate (HR) to assess discrimination of unimodal auditory stimuli (Experiment 1), and to examine how visual stimuli affect auditory discrimination (Experiment 2). The results indicate that the presence of visual stimuli facilitated, rather than interfered with, auditory processing. Keywords: Cognitive Development, Attention, Heart Rate, Psychology, Human Experimentation. Introduction There are many tasks that require people to integrate information across sensory modalities (e.g., associating words with objects and categories, learning the sounds that objects make, etc.). While simultaneously presenting information to different sensory modalities can sometimes facilitate learning, there are also many occasions when presenting stimuli to one modality interfere with learning in a different modality (i.e., modality dominance). Interestingly, the study of modality dominance has generated seemingly inconsistent findings. On the one hand, there is more that 30 years of research on the Colavita effect in adults (Colavita, 1974; Colavita & Weisberg, 1979; Klein, 1977; Posner, Nissen, & Klein, 1976, see Sinnett, Spence, & Soto-Faraco, 2007 for a review). The main finding of these studies is that visual information often interferes with the detection of auditory input, hence the “visual dominance effect”. On the other hand, studies with infants and young children often demonstrate the opposite finding: auditory input often interferes with visual processing, hence the “auditory
Do Children Need Concrete Instantiations to Learn an Abstract Concept? Jennifer A. Kaminski (kaminski.16@osu.edu) Center for Cognitive Science, Ohio State University 210A Ohio Stadium East, 1961 Tuttle Park Place, Columbus, OH 43210, USA Vladimir M. Sloutsky (sloutsky.1@osu.edu) Center for Cognitive Science, Ohio State University 208C Ohio Stadium East, 1961 Tuttle Park Place, Columbus, OH 43210, USA Andrew F. Heckler (heckler.6@osu.edu) Department of Physics, Ohio State University 191 W. Woodruff Avenue, Columbus, OH 43210 USA analogous roles. However, when elements play different roles across domains, transfer typically goes awry (Ross, 1987, Another factor that has been shown to affect transfer is the degree of concreteness of the learning domain (Goldstone & Sakamoto, 2003; Sloutsky, Kaminski, Heckler, 2005). Concrete instantiations communicate more information than their abstract counterparts. For example, consider the increase in information from an abstract stick figure to a detailed drawing, to a photograph. Sometimes this additional information may help communicate the to-be-learned concept and thus concreteness is “Relevant Concreteness”. Other times it may be extraneous, creating “Irrelevant Concreteness”. A recent study examined the effects on learning and transfer of relevant and irrelevant concreteness (Kaminski, Sloutsky, & Heckler, 2005). College undergraduate students learned a simple mathematical concept that was instantiated through different artificial domains. The goal of the study was to investigate whether instantiating an abstract concept in a concrete manner would have benefits or costs for learning and transfer. The type of instantiation learned was a between-subjects factor. Participants learned instantiations that were generic, communicated relevant concreteness, communicated irrelevant concreteness, or communicated both relevant and irrelevant concreteness. For relevant concreteness, the storyline and symbols were designed to help communicate the relevant mathematical structure. Colorful, patterned symbols were used to add extraneous, perceptually engaging irrelevant concreteness. Relevant concreteness was shown to have an advantage for quick learning over irrelevantly concrete or generic instantiations. However with slightly lengthier training, the advantage of relevant concreteness over generic disappeared. Most importantly, both relevant and irrelevant concreteness hindered transfer, while generic instantiations promoted transfer (for the hindering effects of concreteness on transfer see also Goldstone & Sakamoto, 2003 & Sloutsky, Kaminski, Heckler, 2005). The results of this study are striking because they contradict the intuition that facilitating learning will translate Abstract The effects of relevant concreteness on learning and transfer were investigated. Sixth grade students learned artificial instantiations of a simple mathematical concept. Some students were presented with instantiations that communicated concreteness relevant to the to-be-learned concept, while others learned generic instantiations involving abstract symbols. Results suggest that relevant concreteness may have some advantage over generic for learning. However, relevant concreteness hinders transfer of conceptual knowledge to novel isomorphic situations, while generic instantiations promote transfer. Keywords: Cognitive Science; Psychology; Learning; Transfer; Analogical reasoning. Education; Introduction Concrete instantiations are popular tools for teaching abstract concepts in elementary and middle school (see Anderson, Reder, & Simon, 1996; Ball, 1992, for reviews). For example, children learn mathematical concepts such as place value with base ten blocks and fractions with representations of portions of pizza. However there is little empirical evidence of the effectiveness of such material for learning abstract concepts or for any advantage over generic, symbolic representations. Supporting evidence is often anecdotal or limited to demonstrations of knowledge in the learning domain. The goal of learning an abstract concept is not simply knowledge of one instantiation; it is the ability to transfer, or apply conceptual knowledge to a novel isomorphic situation. Successful learning does not necessarily result in successful transfer (e.g. Gick & Holyoak, 1980, 1983). One factor that influences transfer of conceptual knowledge or problem solving strategies is the degree of similarity between the learned domain and the target domain. Superficial similarity between the domains such as storyline can facilitate the retrieval of an analogous, previously learned domain (Gentner, Rattermann, & Forbus, 1993). Also, elements that are similar across domains can promote transfer if they play
