The time course of object encoding

Abstract

Four experiments examined the effects of encoding time on object identification priming and recognition memory. After viewing objects in a priming phase, participants identified objects in a rapid stream of non-object distracters; display times were gradually increased until the objects could be identified (Experiments 1–3). Participants also made old/new recognition judgments about previously viewed objects (Experiment 4). Reliable priming for object identification occurred with 150 ms of encoding and reached a maximum after about 300 ms of encoding time. In contrast, reliable recognition judgments occurred with 75 ms of encoding and continued to improve for encoding times of up to 1200 ms. These results suggest that recognition memory may be based on multiple levels of object representation, from rapidly activated representations of low-level features to semantic knowledge associated with the object. In contrast, priming in this object identification task may be tied specifically to the activation of representations of object shape.

Introduction

Viewing a familiar object activates multiple mental representations. The pattern of light on the retina ultimately activates a hierarchy of visual representations, from those that represent oriented bars (e.g., in primary visual cortex) to those that represent more complex shape features (e.g., in the temporal lobe). These visual representations may ultimately activate semantic information including the name of the object. Given sufficient exposure time, viewing an object may lead to an explicit memory trace that would allow the object to be recognized as having been presented before. In addition, viewing an object may result in implicit memory that is measured by priming. Evidence suggests that the representations that support explicit memory and priming are dissociable (Schacter, Chiu, & Ochsner, 1993).

Explicit memory traces may include almost all the information about the original experience of viewing the object. Thus, in addition to all the specific visual characteristics of the object (e.g., shape, size, left–right orientation, and contrast), the explicit memory may also contain semantic and phonological information as well as episodic information regarding the spatio-temporal context of the item’s presentation. These explicit memory traces can be assessed using a recognition task where one is asked to remember if the object was seen previously (Biederman and Cooper, 1991a, Cooper et al., 1992, Potter, 1976).

Implicit memory traces can also demonstrate the effects of previous experience with objects. For example, objects seen previously are primed in that they can be identified faster or more accurately than objects not seen previously (Bartram, 1974, Biederman and Cooper, 1991a, Biederman and Cooper, 1991b, Biederman and Cooper, 1992, Cooper et al., 1992, Jolicoeur, 1985). Presumably, object priming results from residual activation or an improved efficiency of the neural representations that mediate object identification in the priming (study) exposure. During the subsequent (test) exposure, this residual activation or improved efficiency facilitates object identification.

The representations mediating object priming and explicit object memory have been shown to differ in a number of ways. Object priming appears to depend upon specific visual representations of shape (Biederman and Cooper, 1991a, Biederman and Cooper, 1991b, Biederman and Cooper, 1992, Cooper et al., 1992). For example, an upright piano will prime a grand piano less than the grand piano primes itself, even though they both share semantic and phonological representations. Although these shape representations are specific, they are also abstract in that they do not appear to be tied to very early visual representations -- object shape priming is generally insensitive to changes in the location, size and orientation of the object (Biederman and Cooper, 1991a, Biederman and Cooper, 1991b, Biederman and Cooper, 1992, Cooper et al., 1992, Knowlton et al., 2009, Newell et al., 2005). Explicit memory traces of objects appear to be much more comprehensive. Unlike shape representations mediating identification priming, explicit memory representations may include information about an object’s location, size and left–right orientation (Biederman and Cooper, 1991a, Cooper et al., 1992). In addition, it appears that these explicit memory representations can be accessed in a much more flexible way. For example, participants can easily recognize words referring to pictures that were studied (Nelson and Brooks, 1973, Snodgrass and Asiaghi, 1977), even though the visual form of a word referring to an object is vastly different than the picture of the object.

It appears that different brain structures may be preferentially involved in each type of memory representation. For example, medial temporal lobe damage that occurs in amnesia can impair explicit memory for objects while leaving object priming intact (Cave & Squire, 1992). Conversely, damage to right occipital areas can impair some forms of priming while sparing recognition memory (Wagner, Stebbins, Masciari, Fleischman, & Gabrieli, 1998).

Another potential difference between representations that mediate recognition and those that mediate priming is the time course of encoding these representations. Explicit memory traces may require more encoding time than object priming for a number of reasons. For example, the explicit representations contain more information than simply object shape (e.g., semantic and name information), and these non-shape representations may take more time to fully activate. In addition, it may require additional time to completely bind the object representations to the spatio-temporal context. In contrast, object identification priming may be maximal after a relatively short encoding duration. Object identification is remarkably fast – many familiar objects can be recognized with display times of less than 100 ms (McAulife and Knowlton, 2000, Subramaniam et al., 2000). Thus, unlike recognition, priming may not benefit from encoding times that are much longer than the time required for accurate identification.

In the present study, we wanted to measure the effect of encoding time on object priming and explicit object memory. Importantly, we wanted to measure object priming that was strictly visual; that is, priming for the object shape and not priming of semantic or name information associated with the object. Many previous studies of object identification priming have relied on decreases in naming latency as a measure of priming. This measure includes both visual priming and semantic priming, since accessing the name of an object relies on semantic processing (Biederman and Cooper, 1991a, Biederman and Cooper, 1991b, Biederman and Cooper, 1992, Cooper et al., 1992). The task used in the present set of experiments has been shown to tap into priming that is purely visual, and is not influenced by priming of semantic information (Knowlton et al., 2009, McAuliffe and Hummel, 1998). In the priming phase of this task, subjects view a series of object images. In the probe phase, they view rapid serial visual presentation (RSVP) streams in which an object is included among non-object distracters. The dependent measure is the minimal stimulus onset asynchrony (SOA) required for the object to be identified. Priming is measured as the difference between the minimum identification SOA for old and new objects.

In the first three experiments, we examined the effects of encoding time on object identification priming. Encoding time was defined as the interval from the onset of one object to the next (SOA). In Experiment 1, the encoding time was varied logarithmically in five steps from 300 to 4800 ms (i.e., 300, 600, 1200, 2400 and 4800) with objects displayed for the first 150 ms of the encoding interval. This range corresponds roughly to that used by Potter and Levy (1969) to examine the effects of encoding time on recognition memory. Based on the results of Intraub (1980), we expected that the encoding time manipulation would be the critical influence on object priming even though objects in all conditions were displayed for 150 ms.

The following two experiments were designed to examine the priming resulting from shorter encoding times. In Experiment 2, the encoding time was varied logarithmically in a similar way from 75 to 1200 ms with objects displayed for the first 75 ms of the interval. In Experiment 3, the encoding time was varied from 75 to 300 ms. Importantly, with each experiment, display time was held constant across the varying encoding times. In this way, we were able to measure the effect of encoding time unconfounded with the amount of perceptual information available to the subject.

In Experiment 4, we replicated the encoding phase of Experiment 2, but used an old/new recognition judgment as a probe task to assess explicit memory. Experiment 4 was designed to examine the effects of encoding time on recognition memory using the same materials used in the priming study. We assessed recognition memory using a scale that measured the confidence that the item was seen previously (e.g., 6 = very sure that the object was seen before, 5 = quite sure that the object was seen before, ... 1 = very sure that the object was not seen before). Since identical stimuli and encoding times were used in Experiments 2 and 4, we would be able to directly compare the effects of encoding time on priming and recognition. Based on apparent differences between the representations that support priming and those that support recognition, we predicted that there would be differences in the effect of encoding time between these two measures. Specifically, we hypothesized that recognition memory would benefit from longer encoding times because of the richness of the underlying representation.