www.elsevier.com/locate/ynimg NeuroImage 30 (2006) 1003 – 1009

Category-specific effects in semantic memory: Category–task interactions suggested by fMRI Murray Grossman,a,* Phyllis Koenig,a John Kounios,b Corey McMillan,a Melissa Work,a and Peachie Moore a a

Department of Neurology-2 Gibson, Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104-4283, USA Department of Psychology, Drexel University, Chestnut Street, PA 19104, USA

b

Received 18 August 2004; revised 24 September 2005; accepted 7 October 2005 Available online 18 January 2006

Much work has investigated the neural representation of specific categories of knowledge, but relatively scant attention has been paid in the cognitive neuroscience literature to the semantic processes that contribute to semantic memory. In this study, we monitored regional cortical activity with fMRI while healthy young adults evaluated visually displayed NATURAL KIND, ARTIFACT, and ABSTRACT nouns with two standard tasks: Typicality judgments and Pleasantness judgments. We observed a significant interaction effect between the category of knowledge and the type of judgment used to evaluate members of these semantic categories. Typicality judgments recruited greater temporal – occipital activation relative to Pleasantness judgments of the same category, and this was seen for comparisons of all three semantic categories. However, when contrasted with Typicality judgments, Pleasantness judgments activated a different anatomic distribution for each semantic category. These findings are consistent with a dynamic approach to semantic memory that includes at least two components: semantic knowledge and semantic processes that interpret this knowledge in several ways depending on the particular semantic challenge. D 2005 Elsevier Inc. All rights reserved. Keywords: fMRI; Semantic memory; Temporal

Introduction Recent investigations of semantic memory have focused on hypotheses examining the neuroanatomic representation of category-specific knowledge. Observations of patients with focal cortical insult and functional neuroimaging studies probing semantic memory in neurologically intact adults have revealed dissociations between specific semantic categories. One hypothesis proposes that the localization of a category is an emergent property of the neuroanatomic representation of the features comprising that

* Corresponding author. Fax: +1 215 349 8464. E-mail address: [email protected] (M. Grossman). Available online on ScienceDirect (www.sciencedirect.com). 1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2005.10.046

category. From this perspective, visual – perceptual features may play a particularly prominent role in NATURAL KINDS (we use capitals to indicate a category), and NATURAL KINDS thus may be associated with visual association cortex in ventral temporal and occipital cortical regions where these features are represented (Cappa et al., 1998; Chao et al., 1999; Damasio et al., 1996; Martin et al., 1996; Moore and Price, 1999; Perani et al., 1995; Smith et al., 2001). By comparison, ARTIFACTS may be related more closely to visual motion and action features, and thus are associated with lateral temporal – occipital regions of visual association cortex and motor association cortex in the frontal lobes where visual motion and motor action features may be represented (Cappa et al., 1998; Chao et al., 1999; Grabowski et al., 1998; Martin et al., 1996; Perani et al., 1995). Yet others argue that knowledge of specific categories is represented in a distributed manner (Devlin et al., 2002b; Tyler and Moss, 2001). From this perspective, observations of category-specific dissociations do not result from different neural representations of distinct types of features, but instead are a byproduct of structural differences between categories, such as differences in the magnitude of feature intercorrelations across the exemplars of a category. Attempts to explain category-specific effects solely on the basis of the apparent differences in the type of knowledge associated with each category in semantic memory have encountered considerable difficulty. The neuroimaging literature is quite inconsistent concerning the activations associated with particular categories of knowledge (Farah and Aguirre, 1999; Grossman and Koenig, 2001; Joseph, 2001). Activation for NATURAL KINDS is not restricted to ventral temporal – occipital cortex, for example, but is also associated with lateral temporal cortex (Moore and Prince, 1999) and frontal cortex (Cappa et al., 1998; Grabowski et al., 1998; Martin et al., 1996). Recruitment for ARTIFACTS is not restricted to posterolateral temporal and lateral frontal regions, but is also associated with ventral temporal – occipital cortex (Cappa et al., 1998; Chao et al., 1999; Damasio et al., 1996). Likewise, it is difficult to reconcile claims about the neural representation of sensory-motor features of categories in modality-specific associa-

1004

M. Grossman et al. / NeuroImage 30 (2006) 1003 – 1009

tion cortex when ABSTRACT categories impoverished in sensorymotor features activate these same cortical areas (Beauregard et al., 1997; Grossman et al., 2002a; Kiehl et al., 1999; Kounios et al., 2003; Noppeney and Price, 2002b). In the present study, we tested the hypothesis that some of the inconsistencies in the neuropsychological and neuroimaging literatures are due in part to the flexible manner in which semantic knowledge can be processed. It is possible to understand an object from multiple perspectives. In the context of a meal, a tomato can be sliced and eaten; in the context of a politician, a tomato is a round, squishy object that can be thrown. A trash basket turned upside down can be a chair or a drum. An object with a particular set of perceptual features thus can be categorized in semantically. A long tradition of investigation recognizes the knowledge/process distinction within the information-processing framework (Anderson, 1978; Chang, 1986; Hollan, 1975; Smith et al., 1974). This work highlights the difficulties inherent in deciding whether to attribute particular experimental findings to the nature of the stored semantic knowledge utilized in a particular task, or to the nature of the processing mechanisms that access and transform this activated knowledge. Electrophysiological time-course studies have supported this distinction, for example, by demonstrating that results from a variety of semantic tasks are due to characteristics of the downstream mechanisms that integrate primitive semantic feature knowledge to form higher-level representations and that make decisions based on these assembled knowledge representations rather than being directly determined by the nature of the accessed semantic features (Holcomb et al., 1999; Kounios, 1996; Kounios and Holcomb, 1992). Results from neuroimaging studies of semantic memory likewise show the influence of the tasks used to probe stored semantic knowledge (Farah and Aguirre, 1999). One meta-analysis emphasized this point by examining the neural representation of NATURAL KINDS and ARTIFACT categories of knowledge during the administration of several different tasks (Devlin et al., 2002a). The authors found reliable activation patterns for a specific category of knowledge, but only for tasks involving word retrieval and semantic decision: NATURAL KINDS were associated with bilateral anterior temporal cortex, and ARTIFACTS activated left posterolateral temporal cortex. Other activations were seen during other challenges posed to process these same categories. In another study, a comparison of repetition and semantic decisions about words presented auditorily showed activation of posterior inferior temporal, inferior frontal, and orbital frontal regions of the left hemisphere for both tasks. However, left ventral inferior frontal, left ventral anterior temporal, right cerebellar, and paracingulate regions were recruited only during semantic decisions (Noppeney and Price, 2002a). Recent work has explicitly examined cortical recruitment during the course of evaluating stimuli with different semantic categorization processes (Grossman et al., 2002c; Koenig et al., 2005; Patalano et al., 2001). Rule-based semantic categorization involves the deliberate, feature-by-feature consideration of the features of an ambiguous object with respect to determining its membership in a semantic category, and this appears to entail prefrontal activation. By comparison, similarity-based semantic categorization involves a global comparison of an object with other category members, and this semantic categorization process activates posterolateral temporal – parietal cortex that appears to be important for feature integration. Though intriguing, these studies do not directly examine effects of semantic tasks across categories of knowledge in a

within-subject design, leaving open the possibility that extraneous factors related to individual subject differences may have influenced the results. The present neuroimaging study therefore directly tests the hypothesis of an interaction between task-specific and category-specific effects on patterns of brain activation in semantic memory using a within-subject factorial design that crossed type of task with category of knowledge. Specifically, on different blocks of trials, subjects evaluated words from NATURAL KIND, ARTIFACT, and ABSTRACT semantic categories, and these categories were probed with Typicality judgments or Pleasantness judgments. We selected these judgments for several reasons. First, these two probes have a long history in psychology. Neither involves an objectively correct response, and hence semantic processing is unlikely to be confounded with other task-related components such as goal-oriented problem-solving. The neural basis for these tasks has been investigated in a handful of previous studies. Some work has emphasized the affective nature of ‘‘pleasantness,’’ recruiting a portion of the limbic system in medial frontal cortex (Drevets et al., 1997; Drevets and Raichle, 1998); other work used ‘‘pleasantness’’ to judge affectively neutral categories of knowledge since it can be applied to virtually any semantic category, and showed differential activation depending on the semantic category being probed (Grossman et al., 2002a,b). A ‘‘typicality’’ probe resembles the similarity judgments used in measures of categorization that assess the likeness of a test stimulus to a semantic category’s central tendency or category members (Koenig et al., 2005; Patalano et al., 2001). This work showed activation of temporal – parietal – occipital regions. Finally, although the crucial issue for the purpose of the present study is only that Typicality and Pleasantness are different semantic tasks, there is nevertheless some work suggesting the basis for a difference. A Typicality judgment thus seems to involve a comparison between a stimulus and other members of the same semantic category (Rosch, 1975; Rosch and Mervis, 1975), while a Pleasantness judgment involves the evaluation of a stimulus independent of other objects and concepts (Warrington and Weiskrantz, 1968). In sum, if these tasks evoke different semantic considerations, we would expect to see differential activation for categories of knowledge depending on whether stimuli are being judged during the Typicality task or the Pleasantness task, consistent with a flexible and adaptive characterization of the semantic memory system.

Methods Subjects Nine healthy young subjects (6 females, 3 males) with a mean age of 23.1 years (SD T 3.5 years) and mean education of 16.4 years (SD T 2.3 years) participated in all conditions of this study. All were right-handed native speakers of English. One subject’s behavioral and imaging data for one run (see below) were excluded because of equipment malfunction. A tenth subject was tested but excluded because his behavioral responses suggested that he had not attended to the task: in one condition, for example, he failed to respond to over 22% of the items, exceeding the group mean for missed responses by over two standard deviations; and he produced 11 intrusions (i.e., button press responses in the absence of stimuli), in contrast to the other subjects, none of whom produced such intrusions.

M. Grossman et al. / NeuroImage 30 (2006) 1003 – 1009

Materials and procedures Stimuli were printed nouns comprising three categories: NATURAL KINDS, MANUFACTURED ARTIFACTS, and ABSTRACT terms. The NATURAL KINDS included animals (e.g., monkey, owl) and plants (e.g., clover, apple); the MANUFACTURED ARTIFACTS included manipulable implements (e.g., broom, skillet) and buildings (e.g., bungalow, manor); the ABSTRACT terms included mental states (e.g., dismay, admiration) and legal/political terms (e.g., waiver, allegation).1 There were 120 exemplars from each of the three categories (60 words from each subtype). These words were drawn from a list of 100 of each subtype, which was rated by a cohort of six undergraduate students for familiarity and typicality on scales of 1 to 5. The words used in the study were selected such that each subtype set, and hence each category, contained an equal distribution of items judged to be typical and atypical. Word sets were also equated across categories for familiarity, frequency, and word length. Sixty pronounceable pseudowords (e.g., lorprud, blix) served as base-line stimuli. Subjects saw these stimuli during two probe conditions, Pleasantness and Typicality, as described below. Pleasantness judgments Subjects were instructed to judge each word for ‘‘Pleasantness,’’ and indicate their choices by button-press with a thumb on a hand-held button box (a right-hand button for ‘‘pleasant’’ and a left-hand button for ‘‘not pleasant’’). Responses and reaction times were recorded. Prior to entering the scanner, subjects performed a brief practice task at a laptop computer during which they indicated Pleasantness judgments for words appearing on the screen. Stimuli were presented in a manner blocked by category, but subjects were not explicitly told that the words were grouped by category. Typicality judgments Each block began with the subcategory name (e.g., ANIMALS) appearing for 2 s. Subjects were instructed to judge whether each word was a typical member of its category or not, indicating typicality with a right-hand button press and nontypicality with a left-hand button press. Data collection was otherwise identical to the Pleasantness probe condition. Subjects performed a brief practice task in the scanner prior to the Typicality probe runs. Each probe condition consisted of five runs lasting 7 min and 14 s each. Each run consisted of seven 62-s blocks; runs were separated by a brief period of rest. There were two blocks in each run for each of the three categories (e.g., NATURAL KINDS), and each block contained 12 words from a single subcategory (e.g., animals). An additional block in each run contained pseudowords. Words appeared for 2 s and were presented in a ‘‘jittered’’ pattern. Hence, the stimulus onset asynchronies (SOAs) were 2, 4, 6, and

1 Our focus in this paper was comparisons among three major semantic categories, i.e., NATURAL KINDS, ARTIFACTS, and ABSTRACT terms. To that end, we report activation collapsed across the within-category subtypes (e.g. ANIMALS and PLANTS) that we created for the purpose of making typicality judgments a natural evaluation of a basic level exemplar with respect to a superordinate category. Subcategory distinctions are reported elsewhere.

1005

12 s, and an equal number of these SOAs were presented in a fixed random order for members of each semantic category in each run. Subjects saw the same stimulus items in each task but in different runs. We insured that subjects would not develop any expectation of a particular category or jitter pattern by varying the order of presentation of stimulus content and stimulus timing. Thus, the blocks of word categories appeared in fixed semi-random orders that differed across the runs. The presentation orders of word categories and the jitter patterns differed across conditions. In addition, words were re-randomized such that words that appeared together within a single block for the Pleasantness probe condition did not appear within a single block for the Typicality probe condition. The pseudowords appeared in the fourth (that is, middle) block in all runs in both conditions. Stimuli consisted of printed black lower-case letters in arial font against a white background, and were video-projected onto a screen that subjects viewed via a system of mirrors while in the magnet bore. Presentation was controlled by a computer using the PsyScope software package (Cohen et al., 1993), which also recorded responses. Across conditions, five subjects were presented with the runs in a set random order, while the remaining four were presented with the runs in the opposite order. Imaging procedures and analyses The experiment was performed at 4 T on a GE prototype Echospeed scanner capable of ultrafast imaging. We used a standard clinical quadrature radiofrequency head coil. Firm foam padding was used to restrict head motion. Each imaging protocol began with a 10 – 15 min acquisition of 5 mm adjacent slices for determining regional anatomy, including sagittal localizer images (TR = 500 ms, TE = 10 ms, 192  256 matrix), T2-weighted axial images (FSE, TR = 2000 ms, TEeff = 85 ms), and T1-weighted axial images of slices used for fMRI anatomic localization (TR = 600 ms, TE = 14 ms, 192  256 matrix). Gradient echo echoplanar images were acquired for detection of blood oxygenation level dependent (BOLD) alterations that accompany increased mental activity. All images were acquired with fat saturation, a rectangular FOV of 20  15 cm, flip angle of 30-, 5 mm slice thickness, an effective TE of 50 ms, and a 64  40 matrix, resulting in a voxel size of 3.75  3.75  5 mm. The echoplanar acquisitions consisted of 18 contiguous transaxial slices covering the entire brain every 2 s. Event-related data were acquired with a fixed, randomly ordered stimulus onset asynchrony of 2, 4, 6, or 12 s. This was done in order to selectively average the HRF of each condition over the 12 s time series needed to efficiently measure the HRF (Dale and Buckner, 1997; Friston et al., 1999). There is a risk of biasing parameter estimates, resulting in artificially increased or decreased contrasts, since this TR is a fixed multiple of the stimulus onset asynchrony. A separate acquisition lasting 1 – 2 min was needed for phase maps to minimize distortion in echoplanar images (Alsop, 1995). We also inspected raw data of individual subjects. Raw data were stored by the MRI computer on DAT tape and then processed off-line. Initial data processing was carried out with Interactive Data Language (Research Systems) on a Sun SunBlade 1000 workstation. Raw image data were reconstructed using a 2D FFT with a distortion correction to reduce artifact due to magnetic field inhomogeneities. Individual subject data were then prepared and analyzed statistically using statistical parametric mapping (SPM 99) developed by Wellcome Department of Cognitive Neurology

1006

M. Grossman et al. / NeuroImage 30 (2006) 1003 – 1009

Table 1 Mean (TSD) typicality and Pleasantness judgments and latencies

Pleasantness latency (ms) Typicality latency (ms)

NATURAL KINDS

MANUFACTURED ARTIFACTS

ABSTRACT NOUNS

1034 (223)

1154 (200)

1171 (227)

933 (254)

1102 (209)

1106 (229)

(Frackowiak et al., 1997). This system, operating on a MatLab platform, combines raw difference images from individual subjects into a statistical t score map. We used a random-effects analysis to test our hypotheses. The images in each subject’s time series were registered to the initial image in the series. The images were then aligned to a standard coordinate system (Talairach and Tournoux, 1988). Global perfusion was included as a covariate across groups, the data were smoothed spatially with an 8 mm Gaussian kernel to account for small variations in the location of activation and gyral anatomy across subjects, and low-pass filtering was implemented to control auto-correlation with a first-order auto-regressive method. Each category was contrasted with pronounceable pseudowords that served as a lower level sensory-motor control and to control for drift during the course of the experiment, and the resultant contrasts were investigated further as indicated below. We set a statistical height threshold of the peak voxel in each activated area at P < 0.001 uncorrected, unless otherwise indicated, and an extent threshold at 20 adjacent voxels.

Results Pleasantness and Typicality judgments were evaluated behaviorally by response latencies recorded in ms. Table 1 summarizes Typicality and Pleasantness response latencies for each category of knowledge. Latencies were similar across the ‘‘Pleasantness’’ and ‘‘Typicality’’ types of judgments, suggesting that differences between these probes cannot be attributed to factors like unequal task difficulty. However, we observed more rapid latencies for responses to NATURAL KINDS than to the other categories for both Typicality judgments and Pleasantness judgments. This was confirmed by a repeated-measures ANOVA with a judgment  category design. This showed a main effect for category [F(2,16) = 93.01; P < 0.001]. Paired-sample t tests revealed that response latencies for NATURAL KINDS are more rapid than for

ARTIFACTS [t(6) = 10.30; P < 0.001] and ABSTRACT nouns [t(8) = 18.56; P < 0.001]. We used the response latencies for each category as covariates in the imaging contrasts that compared activations across pairs of semantic categories. We found no differences in activation across semantic categories, so these between-category effects are not considered further. The ANOVA assessing response latencies did not show an effect for judgment [F(1,8) = 3.58; ns], nor was there a judgment  category interaction [ F(2,16) = 0.57; ns]. Table 2 summarizes significant activations suggesting that the distribution of recruitment associated with a category of knowledge depends in part on the nature of the probe. Thus, activations for Typicality judgments invoked a similar pattern of activation for all categories, but activations for Pleasantness judgments varied depending on the semantic category. In direct contrast to Pleasantness judgments of NATURAL KINDS, Fig. 1A shows that Typicality judgments of NATURAL KINDS recruited bilateral temporal – occipital cortex. By comparison, Pleasantness judgments of NATURAL KINDS, directly contrasted with Typicality judgments of NATURAL KINDS, showed bilateral prefrontal and dorsal anterior cingulate recruitment (Fig. 1B). Fig. 1C shows that Typicality judgments of ARTIFACTS, when directly contrasted with Pleasantness judgments of ARTIFACTS, demonstrated bilateral temporal – occipital activation. Pleasantness judgments of ARTIFACTS, directly contrasted with Typicality judgments of ARTIFACTS, showed left temporal – parietal and right ventral temporal recruitment (Fig. 1D). Finally, Fig. 1E illustrates the contrast of Typicality judgments of ABSTRACT nouns minus Pleasantness judgments of ABSTRACT nouns, showing bilateral temporal – occipital recruitment. Relative to Typicality judgments of ABSTRACT nouns, Pleasantness judgments of ABSTRACT nouns recruited right frontal and parietal cortex, as shown in Fig. 1F.

Discussion We observed partially distinct patterns of neural activation associated with each category of knowledge, depending on the nature of the judgment used to probe the category. Typicality judgments elicited a consistent pattern of temporal – occipital activation, emphasizing the role of this anatomic region regardless of the semantic category being judged. By comparison, Pleasantness judgments recruited different anatomic regions for each semantic category, suggesting that the domain of knowledge under consideration contributes to the activation pattern. These observa-

Table 2 Between-judgment contrasts for each category of knowledge Condition

Anatomic locus (Brodmann area)

Coordinates x

NATURAL: Typicality minus Pleasantness NATURAL: Pleasantness minus Typicality ARTIFACTS: Typicality minus Pleasantness ARTIFACTS: Pleasantness minus Typicality ABSTRACT: Typicality minus Pleasantness ABSTRACT: Pleasantness minus Typicality a

Significant at the P < 0.002 level.

Bilateral temporal – occipital (37, 17, 18) Left prefrontal (6, 8) Right prefrontal (6, 8) Right temporal – occipital (37, 18, 19) Left temporal – occipital (19) Left lateral temporal – parietal (39, 40) Right medial ventral temporal (28, 36) Bilateral temporal – occipital (37, 17, 18) Right lateral parietal (40) Right prefrontal (6, 8)

y 8 28 28 12 24 55 8 4 51 40

# voxels

z score

418 465 465 253 60 48 20 423 21 45

3.84 3.65 3.73 4.10 2.89a 3.40 3.15 3.20 3.67 2.92a

z 81 14 14 76 76 56 24 85 48 6

4 44 44 26 33 36 16 4 43 40

M. Grossman et al. / NeuroImage 30 (2006) 1003 – 1009

1007

Fig. 1. Between-judgment contrasts each category of knowledge. (A) Typicality judgments minus Pleasantness judgments for NATURAL KINDS; (B) Pleasantness judgments minus Typicality judgments for NATURAL KINDS; (C) Typicality judgments minus Pleasantness judgments for ARTIFACTS; (D) Pleasantness judgments minus Typicality judgments for ARTIFACTS; (E) Typicality judgments minus Pleasantness judgments for ABSTRACT nouns; (F) Pleasantness judgments minus Typicality judgments for ABSTRACT nouns.

tions are consistent with the hypothesis that semantic memory involves at least two components, including semantic knowledge representations and the processes used to interpret this knowledge. In the present study, we probed the same categories of knowledge in a single group of subjects with two different semantic judgments, both of which involve a dichotomous decision and the same motor response about the same words. Analyses of subjects’ behavioral responses revealed a difference between the category of semantic knowledge being judged. Subjects were more rapid at judging NATURAL KINDS than ARTIFACTS or ABSTRACT nouns. This has been described previously (Flores d’Arcais and Schreuder, 1987; Flores d’Arcais et al., 1985). We covaried for the response latencies associated with each semantic category during contrasts of the associated activations, but no between-category differences were evident. This negative finding should be interpreted very cautiously because of the small number of participants. Nevertheless, other reports have described activa-

tion patterns that are similar across different semantic categories (Devlin et al., 2002b; Tyler et al., 2000, 2003). This finding has been interpreted to support the claim that the neural representation of semantic knowledge is distributed in nature. It is important to point out that the similarity of activation patterns across semantic categories in the present study depends on the judgment used to probe these categories of knowledge. Thus, we observed distinct patterns of activation for NATURAL KIND, ARTIFACT, and ABSTRACT nouns depending on whether knowledge of these categories was probed with Typicality judgments or Pleasantness judgments. Direct contrasts of these probes for each semantic category revealed greater temporal – occipital cortex recruitment for Typicality judgments than Pleasantness judgments. Although we cannot describe the precise basis for activations associated with these probes, it is not unreasonable to speculate that Typicality judgments may have encouraged subjects to assess the resemblance of a test stimulus to a prototype or to

1008

M. Grossman et al. / NeuroImage 30 (2006) 1003 – 1009

other members of a category (Rosch, 1975; Rosch et al., 1976). From this perspective, Typicality judgments of any category result in activation that includes modality-specific association regions important for evaluations of attributes of semantic categories contributing to these comparative judgments. One possibility is that this is related to the activation of sensory-motor feature knowledge associated with the semantic representation of an object category, as in the case of NATURAL KINDS and ARTIFACTS. This distribution of activation has been seen commonly for NATURAL KINDS and ARTIFACTS (Cappa et al., 1998; Chao et al., 1999; Damasio et al., 1996; Martin et al., 1996; Moore and Prince, 1999; Perani et al., 1995; Smith et al., 2001). Since the same activation was demonstrated for ABSTRACT nouns that have few perceptual features, an alternate account emphasizes the comparison process underlying Typicality judgments (Koenig et al., 2005; Patalano et al., 2001). This process thus may have activated a distributed semantic network not directly associated with specific sensory-motor features that nevertheless contributes to the performance of a Typicality judgment. When we examined the activation patterns for Pleasantness judgments relative to Typicality judgments, by comparison, we observed recruitment that varies depending on the NATURAL KIND, ARTIFACT, and ABSTRACT semantic category being probed. This parallels previous findings that used Pleasantness judgments to probe these semantic categories (Grossman et al., 2002a). Pleasantness judgments resulted in greater activation of bilateral prefrontal regions during probes of NATURAL KINDS compared to Typicality judgments. Previous work has shown frontal activation during probes of NATURAL KINDS (Cappa et al., 1998; Grabowski et al., 1998; Martin et al., 1996). We may not have seen temporal – occipital activation during the contrast of Pleasantness judgments minus Typicality judgments since temporal – occipital recruitment was present during both Typicality and Pleasantness probes. For ARTIFACTS, Pleasantness judgments resulted in greater left temporal – parietal and right temporal activation than for Typicality judgments. These temporal regions have been associated with ARTIFACTS in previous activation studies (Cappa et al., 1998; Chao et al., 1999; Damasio et al., 1996; Grabowski et al., 1998; Martin et al., 1996; Perani et al., 1995). ABSTRACT nouns recruited temporal and frontal regions of the right hemisphere during Pleasantness judgments relative to Typicality judgments. While some work has emphasized that verbal coding of ABSTRACT nouns should result in left hemisphere activation (Binder et al., 2005), others have emphasized the coarse coding of ABSTRACT concepts that is more likely to evoke right hemisphere activation (Grossman et al., 2002a). Regardless of the basis for the specific activations associated with each category of knowledge, we emphasize here that Pleasantness judgments resulted in category-specific recruitment patterns that were not seen for Typicality judgments. Pleasantness judgments appear to depend less on a comparison process, and instead involve a consideration of the particular features of the concept under consideration. This may have allowed category-specific differences to emerge in the present study and in previous studies using Pleasantness to probe semantic memory and other domains of knowledge (Drevets et al., 1997; Drevets and Raichle, 1998; Grossman et al., 2002a,b). The finding that activations depend in part on the semantic memory probe and in part on the semantic category suggests a view of semantic memory that includes at least two components: the semantic knowledge represented in semantic memory, and the

processes integrating these features for the purpose of solving a semantic challenge. While much previous work has emphasized the representation of semantic knowledge, the view supported by the present study underlines the importance of considering the processes within semantic memory that contribute to the comprehension of noun concepts. We hypothesize that our findings reflect the different ways in which semantic feature knowledge may be integrated to solve a semantic challenge. Several semantic processes have been examined in previous studies of semantic memory. For example, posterolateral temporal activation was observed during several tasks requiring integration of visual and auditory feature knowledge for the purpose of object identification (Beauchamp et al., 2004). Another approach examined semantic categorization of object descriptions (e.g., ‘‘A round object 3 in. in diameter: PIZZA or QUARTER’’) under two conditions: a rule-based condition that required subjects to focus on a specific feature of the object description; or a similarity-based condition that encouraged subjects to evaluate the overall characteristics of the object description during categorization (Grossman et al., 2002c). The findings showed a shift in cortical activation depending on the categorization probe. A follow-up study taught a novel animal to participants in one of two different ways: by a rule-based process involving the deliberate, feature-by-feature consideration of the critical properties that determine its semantic category membership; or by a similaritybased process involving a relatively rapid, global comparison of an object with other category members (Koenig et al., 2005). During subsequent category membership judgments of category members and foils, the results showed different activation patterns depending on the process used to learn the new category. Several caveats must be kept in mind when considering our findings. We examined only a restricted set of semantic categories with a limited range of semantic probes, and additional work is needed to assess the generalizability of our findings. NATURAL KIND, ARTIFACT, and ABSTRACT noun concepts are very complex, and it would be valuable to examine categories with a simpler collection of features. The work in the present study focuses on the categorization processes that are intrinsic to the semantic memory system, but other processes contributing to semantic memory such as selection and retrieval cannot be excluded from any comprehensive consideration of semantic memory. With these caveats in mind, the present results are consistent with a two-component model of semantic memory: semantic knowledge, and semantic categorization processes that construct a meaningful concept from this knowledge.

Acknowledgments This work was supported in part by the US Public Health Service (AG15116, NS44266, AG17586).

References Alsop, D., 1995. Correction of ghost artifacts and distortion in echoplanar MR with an iterative reconstruction technique. Radiology 194, 338. Anderson, J.R., 1978. Arguments concerning representations for mental imagery. Psychol. Rev. 85, 249 – 277. Beauchamp, M.S., Lee, K.E., Argall, B.D., Martin, A., 2004. Integration of auditory and visual information about objects in superior temporal sulcus. Neuron 41, 809 – 823.

M. Grossman et al. / NeuroImage 30 (2006) 1003 – 1009 Beauregard, M., Chertkow, H., Bub, D., Murtha, S., Dixon, R., Evans, A., 1997. The neural substrate for concrete, abstract, and emotional word lexica: a positron emission tomography study. J. Cogn. Neurosci. 9, 441 – 461. Binder, J.R., Westbury, C., McKiernan, K.A., Possing, E.T., Medler, D.A., 2005. Distinct brain systems for processing concrete and abstract concepts. J. Cogn. Neurosci. 905, 917. Cappa, S.F., Perani, D., Schnur, T., Tettamanti, M., Fazio, F., 1998. The effects of semantic category and knowledge type on lexical-semantic access: a PET study. NeuroImage 8, 350 – 359. Chang, T.M., 1986. Semantic memory: facts and models. Psychol. Bull. 99, 199 – 220. Chao, L.L., Haxby, J., Martin, A., 1999. Attribute-based neural substrates in temporal cortex for perceiving and knowing about objects. Nat. Neurosci. 2, 913 – 919. Cohen, J.D., MacWhinney, B., Flatt, M.R., Provost, J., 1993. PsyScope: a new graphic interactive environment for designing psychology experiments. Behav. Res. Methods Instrum. Comput. 25, 101 – 113. Dale, A.M., Buckner, R.L., 1997. Selective averaging of rapidly presented individual trials using fMRI. Hum. Brain Mapp. 5, 329 – 340. Damasio, H., Grabowski, T.J., Tranel, D., Hichwa, R.D., Damasio, A.R., 1996. A neural basis for lexical retrieval. Nature 380, 499 – 505. Devlin, J.T., Moore, C.J., Mummery, C.J., Gorno-Tempini, M., Phillips, J.A., Noppeney, U., et al., 2002a. Anatomic constraints on cognitive theories of cognitive specificity. NeuroImage 15, 675 – 685. Devlin, J.T., Russell, R.P., Davis, M.H., Price, C.J., Moss, H.E., Fadili, M.J., et al., 2002b. Is there an anatomic basis for category-specificity? Semantic memory studies in PET and fMRI. Neuropsychologia 40, 54 – 75. Drevets, W.C., Raichle, M., 1998. Reciprocal suppression of regional cerebral blood flow during emotional versus cognitive processes: implications for interactions between emotion and cognition. Cogn. Emot. 12, 353 – 385. Drevets, W.C., Price, J.L., Simpson Jr., J.R., Todd, R.D., Reich, T., Vannier, M., et al., 1997. Subgenual prefrontal cortex abnormalities in mood disorders. Nature 386, 824 – 827. Farah, M., Aguirre, G., 1999. PET and fMRI studies of the functional anatomy of human visual recognition. Trends Cogn. Sci. 3, 179 – 186. Flores d’Arcais, G.B., Schreuder, R., 1987. Semantic activation during object naming. Psychol. Res. 49, 153 – 159. Flores d’Arcais, G.B., Schreuder, R., Glazenborg, G., 1985. Semantic activation during recognition of referential words. Psychol. Res. 47, 39 – 49. Frackowiak, R.S.J., Friston, K.J., Frith, C.D., Dolan, R.J., Mazziotta, J.C., 1997. Human Brain Function. Academic Press, San Diego. Friston, K., Zarahn, E., Josephs, O., Henson, R.N.A., Dale, A.M., 1999. Stochastic designs in event-related fMRI. NeuroImage 10, 607 – 619. Grabowski, T.J., Damasio, H., Damasio, A.R., 1998. Premotor and prefrontal correlates of category-related lexical retrieval. NeuroImage 7, 232 – 243. Grossman, M., Koenig, P., 2001. Semantic memory. In: Ramachandran, V.S. (Ed.), Encyclopedia of Cognitive Science. Academic Press, San Diego. Grossman, M., Koenig, P., DeVita, C., Glosser, G., Alsop, D., Detre, J., et al., 2002a. The neural basis for category-specific knowledge: an fMRI study. NeuroImage 15, 936 – 948. Grossman, M., Koenig, P., DeVita, C., Glosser, G., Alsop, D., Detre, J., et al., 2002b. The neural representation of verb meaning: an fMRI study. Hum. Brain Mapp. 15, 224 – 234. Grossman, M., Smith, E.E., Koenig, P., Glosser, G., DeVita, C., Moore, P., et al., 2002c. The neural basis for categorization in semantic memory. NeuroImage 17, 1549 – 1561. Holcomb, P.J., Kounios, J., Anderson, J.R., West, W.C., 1999. Dual coding, context availability, and concreteness effects in sentence comprehension: an electrophysiological investigation. J. Exper. Psychol., Learn., Mem., Cogn. 20, 804 – 823.

1009

Hollan, J.D., 1975. Features and semantic memory: set-theoretic or network model? Psychol. Rev. 82, 154 – 155. Joseph, J.E., 2001. Functional neuroimaging studies of category specificity in object recognition: a critical review and meta-analysis. Cogn. Affect. Behav. Neurosci. 1, 119 – 136. Kiehl, K.A., Liddle, P.F., Smith, A.M., Mendrek, A., Forster, B.B., Hare, R.D., 1999. Neural pathways involved in the processing of concrete and abstract words. Hum. Brain Mapp. 7, 225 – 233. Koenig, P., Smith, E.E., Glosser, G., DeVita, C., Moore, P., McMillan, C., et al., 2005. The neural basis for novel semantic categorization. NeuroImage 24, 369 – 383. Kounios, J., 1996. On the continuity of thought and the representation of knowledge: electrophysiological and behavioral time-course measures reveal levels of structure in semantic memory. Psychon. Bull. Rev. 3, 265 – 286. Kounios, J., Holcomb, P.J., 1992. Structure and process in semantic memory: evidence from event-related brain potentials and reaction times. J. Exp. Psychol. Gen. 121, 459 – 479. Kounios, J., Koenig, P., Glosser, G., DeVita, C., Moore, P., Dennis, K., et al., 2003. Category-specific medial temporal lobe activation and the consolidation of semantic memory: evidence from fMRI. Cogn. Brain Res. 17, 484 – 494. Martin, A., Wiggs, C.L., Ungerleider, L.G., Haxby, J.V., 1996. Neural correlates of category-specific knowledge. Nature 379, 649 – 652. Moore, C.J., Price, C.J., 1999. A functional neuroimaging study of the variables that generate category-specific object processing differences. Brain 122, 943 – 962. Noppeney, U., Price, C.J., 2002a. A PET study of stimulus- and taskinduced semantic processing. NeuroImage 15, 927 – 935. Noppeney, U., Price, C.J., 2002b. Retrieval of visual, auditory, and abstract semantics. NeuroImage 15, 917 – 926. Patalano, A., Smith, E.E., Jonides, J., Koeppe, R., 2001. PET evidence for multiple strategies of categorization. Cogn. Affect. Behav. Neurosci. 1, 360 – 370. Perani, D., Cappa, S.F., Bettinardi, V., Bressi, S., Gorno-Tempini, M., Matarrese, M., et al., 1995. Different neural systems for the recognition of animals and man-made tools. NeuroReport 6, 1637 – 1641. Rosch, E., 1975. Cognitive representation of semantic categories. J. Exp. Psychol. Gen. 104, 192 – 233. Rosch, E., Mervis, C.B., 1975. Family resemblances: studies in the internal structure of categories. Cogn. Psychol. 3, 382 – 439. Rosch, E., Mervis, C.B., Gray, W.D., Johnson, D.M., Boyes-Braem, P., 1976. Basic objects in natural categories. Cogn. Psychol. 8, 382 – 439. Smith, E.E., Shoben, E.J., Rips, L.J., 1974. Structure and process in semantic memory: a featural model for semantic decisions. Psychol. Rev. 81, 214 – 241. Smith, C.D., Anderson, A.H., Kryscio, R.J., Schmitt, F.A., Kindy, M.S., Blonder, L.X., et al., 2001. Differences in functional magnetic resonance imaging activation by category in a visual confrontation naming task. J. Neuroimaging 11, 165 – 170. Talairach, J., Tournoux, P., 1988. Co-Planar Stereotaxic Atlas of the Human Brain, 1st edR Thieme Medical Publishing Company, New York. Tyler, L.K., Moss, H.E., 2001. Towards a distributed account of conceptual knowledge. Trends Cogn. Sci. 5, 244 – 252. Tyler, L.K., Moss, H.E., Durrant-Peatfield, M., Levy, J.P., 2000. Conceptual structure and the structure of concepts: a distributed account of category-specific deficits. Brain Lang. 75, 195 – 231. Tyler, L.K., Bright, P., Dick, E., Tavares, P., Pilgrim, L., Fletcher, P., et al., 2003. Do semantic categories activate distinct cortical regions? Evidence for a distributed neural semantic system. Cogn. Neuropsychol. 20, 541 – 559. Warrington, E.K., Weiskrantz, L., 1968. New method of testing longterm retention with special reference to amnesic patients. Nature 217, 972 – 974.

Category-specific effects in semantic memory: Category ...

brief practice task at a laptop computer during which they indicated Pleasantness ... began with a 10–15 min acquisition of 5 mm adjacent slices for determining ...

283KB Sizes 0 Downloads 288 Views

Recommend Documents

Category-specific effects in semantic memory
One meta-analysis empha- ... behavioral and imaging data for one run (see below) were excluded because .... We also inspected raw data of individual subjects.

The spacing effect in children's memory and category ... - CiteSeerX
allowing children time to forget the instances of the cate- gories that they were .... Education, University Parent's Nursery, and Westwood. United Methodist ...

Application-Specific Memory Management in ... - Semantic Scholar
The Stata Center, 32 Vassar Street, Cambridge, Massachusetts 02139. Computer Science and ... ware mechanism, which we call column caching. Column caching ..... in part by the Advanced Research Projects Agency of the. Department of ...

Semantic memory
formal computational models, neural organization, and future directions. 1.1. ... Tulving's classic view of semantic memory as an amodal symbolic store has been ...

Entrenchment & Memory Development Effects on ... -
needs to divide the phoneme sequence first into morphemes and words and, at a higher level, into noun phrases in which gender agreement must be ...

Learning and memory in mimicry: II. Do we ... - Semantic Scholar
Article ID: bijl.1998.0310, available online at http://www.idealibrary.com on ... 1Department of Genetics, University of Leeds, Leeds LS1 9JT ... to the degree of pleasantness or unpleasantness of a prey generates non-monotonic results.

Learning and memory in mimicry: II. Do we ... - Semantic Scholar
We focus on the general dynamics of predator learning and memory. .... post-attack value (before the application of the forgetting routine) by a constant ...... intensive study of the Ithomiine mimicry rings in Amazonian Ecuador, Beccaloni.

Memory in the Occurrence of Earthquakes - Semantic Scholar
Nov 10, 2005 - of the recurrence times in California and obtained a unified scaling law for the spatiotemporal set of data (see also. [13,14]). Corral [10–12] ...

Individual Differences in Psychotic Effects of ... - Semantic Scholar
Jun 18, 2008 - ... Cambridge CB2 2QQ, UK. E-mail: [email protected]. ... tence in either their own voice or one of two robotic voices. Before the study, samples ...

An Energy-efficient Nonvolatile In-memory ... - Semantic Scholar
THE analysis of big-data at exascale (1018 bytes/s or flops) has introduced the emerging need to reexamine the existing hardware platform that can support ...

An Energy-efficient Nonvolatile In-memory ... - Semantic Scholar
to a logic-in-memory architecture by non-volatile domain-wall nanowire. Domain-wall ..... as the magnetization reversal of MTJ free layer, i.e. the target domain of the ...... at 100µA, and the current density at 6 × 108A/cm2 for shift- operation.

Sources of individual differences in working memory - Semantic Scholar
Even in basic attention and memory tasks ... that part-list cuing is a case of retrieval-induced forgetting ... psychology courses at Florida State University participated in partial ... words were presented at a 2.5-sec rate, in the center of a comp

Polarization effects and charge transfer in the KcsA ... - Semantic Scholar
b International School for Advanced Studies, SISSA-ISAS and INFM-Democritos Center, via Beirut 4, 34014 Trieste, ... The electronic structure of the selectivity filter of KcsA K+ channel is ... which features the conserved TVGYG signature [6,7].

Fate and effects of xanthates in laboratory ... - Semantic Scholar
Mean room temperature was 25 C. Losses of water from the nutrient solutions were ... mlned(a] from the sterilized nutrient solution and(bi from the water in the ...

Syllabic length effects in visual word recognition ... - Semantic Scholar
Previous studies on the syllable-length effect in visual word recognition and naming ..... as subjects for course credit, 16 in the Experiment 1 (naming) and 40 in ... presented in isolation on the center of the display screen of a Pentium computer.

Person Memory and Judgment: Pragmatic ... - Semantic Scholar
ert S. Wyer, Jr., Department of Psychology, University of Illinois, 603. East Daniel, Champaign ... These data suggest that subjects with the goal of forming an impression of a person on the basis ..... was not open to question. Moreover, the other .

Person Memory and Judgment: Pragmatic ... - Semantic Scholar
San. Diego, CA: Academic Press. Jones, E. E., Schwartz, J., & Gilbert, D. T. (1984). Perception of moral expectancy violations: The role of expectancy source.

Identifying Social Learning Effects - Semantic Scholar
Feb 11, 2010 - treatment by police officers (often measured as stop or search rates) can ... racial prejudice using a ranking condition that compares searches ...

Identifying Social Learning Effects - Semantic Scholar
Feb 11, 2010 - Our analysis permits unobservables to play a more general role in that we ...... In other words, race has no marginal predictive value for guilt or.

A COHERENT HOMOTOPY CATEGORY OF 2 ... - Semantic Scholar
Subject classifications : [2000] 18D05, 18B30, 55P10. Keywords : track, semitrack, homotopy 2-groupoid, triple category, homotopy pair, interchange 2-track, Toda bracket. Department of Mathematics and Applied Mathematics, University of Cape Town,. 77

the category-partition method for specifying and ... - Semantic Scholar
A method for creating functional test suites has been developed in zohich a test engineer ... Conference,. Portland, .... ample, one possible way to partition the category array size int.o .... contains it (in the test specification, we call such a.

Making working memory work: The effects of extended ...
Jan 29, 2014 - attention for the forward-access task, but not for variations where probing was in random order. This suggests that working memory capacity may depend on the type of search process engaged, and that certain working- memory-related cogn