Search

In a previous study we tested whether a conceptually meaningful object dimension (color) affected memory for another, arbitrary, non-diagnostic dimension/feature (location). We found that, overall, participants exhibited better memory for an arbitrary location of an object that possessed a conceptually meaningful than a meaningless color. Since our study examined the inter-relations among different object dimensions, it may also have important implications for the debate concerning the level of dependency among features in visual LTM of real-world objects (e.g., Balaban et al., 2019; Brady et al., 2013; Utochkin & Brady, 2019). In order to further explore this topic, the present study will include an additional delayed memory test that will be conducted three days after the encoding and the immediate memory test phase. We will explore the effect of time passage on the response pattern, in an attempt to better understand the nature of the relations (in/dependency) between color and location.

In a previous study we found enhanced long-term memory for an object’s color when it was associated with the object’s identity, that is, when the color was diagnostic of an item’s meaning and its conceptual properties. Presumably, color attributes such as a ‘red wine’ or a ‘green apple’ are not merely perceived as surface features, but rather, they are perceived as inherent qualities of an object that are associated with its taste, texture and perhaps other properties. Accordingly, they are better encoded and remembered than colors that are perceived as arbitrary, non-diagnostic features (see also Konkle et al., 2010). In a follow-up study, we further examined the potential influence of a meaningful/diagnostic color on LTM of an independent object property, i.e., its location. That is, we tested whether one (conceptually meaningful) dimension affected memory for another (arbitrary, non-diagnostic) dimension/feature. We found that participants indeed exhibited better memory for an arbitrary location of an object that possessed a conceptually meaningful color. These results support the view that preexisting (conceptual) knowledge may provide a ‘conceptual hook’ for perceptual detail (Brady et al., 2011; Konkle et al., 2010), and/or may free attentional resources required for the encoding of additional item information (Resource-based approach, e.g., Reder et al., 2013, 2016). Importantly, although we aimed to test visual LTM, memory performance related to the color dimension could potentially be accounted for by verbal factors. That is, subtype objects possessing conceptually-meaningful colors may be better encoded verbally than subtype objects which are color-meaningless, since color may be inherent to the name/label of the former (e.g., 'RED WINE'). In addition, different subtypes of color-meaningful objects may carry different names (e.g., 'LIME' vs, 'LEMON'), allowing a more distinctive verbal encoding and, consequently, improved subsequent memory retrieval. Indeed, conceptual information is often closely tied to verbal representations. To exclude 'pure' verbal factors that may potentially assist in visual (and conceptual) memory encoding, the present study will therefore replicate our previous experiment, with the addition of a verbal suppression task during the encoding phase. Specifically, participants will be required to repeatedly count out loud from 1 to 3 in a fixed tempo, while encoding the viewed object targets.

Previous findings have revealed that visual long-term memory can store a massive number of detailed objects (Brady, Konkle, Alvarez, & Oliva, 2008). Using a 2-alternative-forced-choice (2AFC) test that required fine memory distinctions among different exemplars, or different states of the same objects, participants showed a remarkably high memory performance (~ 87%) for hundreds of real-world objects. Surprisingly, however, a pilot conducted in our lab using a similar paradigm, in which participants were specifically tested on memory for an object’s color, revealed rather low memory rates (~67%). These memory rates were obtained despite the fact that the colors of the target (old) and foil (new) items in the 2AFC test were highly distinct from each other (e.g., a blue vs. red balloon). In addition, the encoding phase included less than one eighth of the stimuli used in the Brady et al. (2008) study. That is, despite the rather low encoding load, and a high perceptual distinctiveness among items during the memory test, participants exhibited low memory performance, suggesting that color per-se may not be a highly memorable feature. A possible account for these results is that the object colors in our pilot were determined randomly, such that they were strictly arbitrary and bore no meaning with respect to the tested items. A blue balloon, for instance, is not different in its essence from a red balloon, and the object’s color is superficial to the item’s nature. In contrast, different object exemplars (e.g., different types of guitars) or different object states (an open vs. closed dresser), as the ones used in the Brady et al. (2008) study, may differ qualitatively from each other (i.e., both visually and semantically). To the extent that color affects the perceived meaning, or the perceived ‘essence’ of an object, high memory rates may be obtained when testing visual memory for color detail. If, however, color (or any other surface feature) remains insignificant, or independent of an object’s nature, it may be poorly encoded and remembered. Put differently, the contribution of a feature to the perception and understanding of an object’s essence, or its core meaning, may determine the extent to which it is subsequently remembered. Indeed, LTM is long known to be ‘hooked’ on semantic meaning The present study will examine the role of a color’s meaning to LTM of an object’s visual (color) details. Colors’ meaning will be determined by an online survey, and will serve as a predictor for visual LTM fidelity level (see below).

Wolfe et al. have shown that reaction times in hybrid search tasks are a logarithmic function of the number of items being held in memory. This pattern is highly robust across various conditions. Here, we test whether this logarithmic compression will hold when the search requires finer discrimination between items in memory.

Is fine discrimination in memory (such as discrimination between different states of a same object, or different exemplars of the same basic-level category) characterized by a “logarithmic” search pattern through the memory list in both pure memory search (which is performed under the visual set size of 1 and, hence, does not require visual search) and in a hybrid search? The current experiment will aim to test the character of search through memory alone as a function of memory set size and required memory resolution. This experiment is a prerequisite for future hybrid search experiments.

Visual and memory search requiring low vs. high resolution: Novelty and Familiarity Recognition (Memory Search) as a Function of Memory Set Size

Wolfe et al. (2015, JEP HPP) have shown that the reaction time as a logarithmic function of memory set size in hybrid search tasks is robust when the task requires finding a novel target among familiar distractors. Here, we use this novelty search task to figure out whether this logarithmic compression will hold when the search requires finer discrimination between items in memory.

This is a follow-up experiment to a prior project of Hybrid Partition (Experiments 1&2).

It has been proposed that although unitization benefits associative episodic memory, such benefits may exact a cost of poorer subsequent memory for the individual component elements forming these associations. However, unitization effects on item memory may be a function of whether items were unitized in a top-down fashion (i.e., through encoding instructions) or in a bottom-up fashion (through inherent semantic relationships between the associated items). We hypothesized that semantic relatedness unitization would benefit item recognition due to processing fluency that affords availability of more encoding resources. In two experiments, we tested how relatedness unitization of object pictures at encoding affected subsequent item recognition. The first experiment used a between-subject inclusion/exclusion paradigm, while the second experiment used a within-subject three-alternative choice recognition test. Results of both experiments revealed that semantic relatedness unitization benefited item memory. However, both accuracy patterns and error analyses indicated that such relatedness-based unitization improved memory for an item’s gist (i.e., its categorical identity), but not for its visual details. This indicates that that associative relatedness at encoding strengthens representations of the semantic identity of individual stimuli, but not necessarily of their perceptual details, possibly because of enhanced semantic elaboration at encoding.