1. PET Count  Word Frequency effects (coefficients) were reliably related to activation in both the striate and ITG for older adults only.  For older adults, greater activation in both the striate and ITG was associated with larger effects of frequency on lexical decision RT (i.e., larger frequency coefficients). Previous studies of word identification have found that the contribution of lexical/semantic and sublexical/nonsemantic processing to word identification is different for younger and older adults (e.g., Allen et al., 1991; Spieler & Balota, 2000). Spieler and Balota, like others, found that the more frequently a word appears in printed text, the more rapidly both younger and older adults were able to name words. However, this lexical variable (word frequency) accounted for a greater proportion of variance in older adults’ naming latencies relative to younger adults’. This finding along with other findings of equivalent word frequency effects for younger and older adults during word identification provides evidence of preserved semantic processing with age. There is also evidence to suggest that the relationship between sublexical word properties (e.g., word length) and word identification RT also varies with age (Allen et al.). The goal of the present study was to determine whether age differences in the effects of lexical and sublexical processes on word identification might reflect different patterns of neural activation. Madden et al. (in press) previously examined the relationship between word identification RT and activation, but found no age differences in the correlation between these two variables. The present study reexamined the Madden et al. data to see if age differences in the relationship between RT and activation might be mediated by lexical and sublexical word properties.  We next examined whether activation in Madden et al. might be related to these lexical and sublexical variables.  Madden et al. found that, relative to the baseline search task, word identification was associated with greater activation of the inferior temporal gyrus (ITG) for older adults, but greater visual striate activation for younger adults (see Figure 2 below). Our analyses focused on these 2 areas of activation.  We simultaneously regressed activation (PET counts) in these 2 areas onto word frequency and length coefficients in separate analyses for each lexical/sublexical variable and age group (see Figures 3 & 4).  Word frequency had a larger effect on RT than length for both younger (  = -.497*,R 2 =.168 vs.  =.101*, R 2 =.007) and older adults (  = -.457*, R 2 =.146 vs.  =.241*, R 2 =.041).  To determine whether the pattern of frequency and length effects was statistically different for younger and older adults, we used frequency and length values to predict RTs for each participant. Each participant’s frequency and length coefficient was submitted to an Age x Predictor ANOVA. Lexical and Sublexical Components of Age-Related Changes in Neural Activation During Visual Word Identification Introduction Method  For older adults, greater activation in the striate and ITG was associated with less efficient lexical (word frequency) processing, as those with larger coefficients had slower RTs.  The opposite effect was found for word length: lower striate activation was associated with less efficient sublexical/perceptual (word length) processing.  This pattern of effects may indicate that older adults compensate for early visual processing deficiencies during word identification by engaging in more semantic processing hosted by the inferior temporal gyrus. Conclusions Participants  12 younger adults (M = 23.6 yrs, range 20-29 yrs) and 12 older adults (M = 65.0 yrs, range 62-70 yrs). Design and Procedure Word Identification task:  A standard yes/no lexical decision task: during each trial block, participants were randomly presented with either words (e.g., FOREST) or pronounceable nonwords (e.g., TWEAL).  On a single trial, participants pressed one of two buttons on a response box to indicate whether the letter string was a word or nonword.  Participants performed 9 blocks of trials containing a total of 720 words and nonwords.  Words Ranged in Frequency from 20 to 65 SFI and 4 to 10 letters in length within each block. Baseline task:  For the baseline/subtraction task, participants searched letter strings of “T”s and “Z”s in search of a lower case “c” (e.g., TZTZTZcT) and pressed one of two buttons to indicate the presence or absence of a “c”. Positron Emission Tomography (PET):  Measurement of regional cerebral blood flow (rCBF) was conducted with a GE Advance whole-body PET scanner (35 imaging planes separated by 4.25 mm). Intrinsic in-plane and axial spatial resolutions were approximately 5 mm.  Radiotracer administration was intravenous bolus injection of 10 mCi of H 2 15 O. Data acquisition was performed in the 3D mode (septa out). Each task condition was performed during a separate PET scan. There was a total of 12 emission scans, three scans for each of four task conditions (1 baseline, 3 lexical decision). Spieler, D. H., & Balota, D. A. (2000). Factors influencing word naming in younger and older adults. Psychology and Aging, 15, 225-231. Acknowledgements This work was supported by National Institute on Aging grants: T32 AG00029-26 and R01 AG11622. References Allen, P. A., Madden, D. J., & Crozier, L. C. (1991). Adult age differences in letter-level and word- level processing. Psychology and Aging, 6, 261-271. Madden, D. J., Langley, L. K., Denny, L. L., Turkington, T. G., Provenzale, J. M., Hawk, T. C., et al. (in press). Adult age differences in visual word identification: Functional neuroanatomy by positron emission tomography. Brain & Cognition. Behavioral Results Activation Results Whiting, W.L., Madden, D.J., Langley, L.K., Denny, L.L., Turkington, T.G., Provenzale, J.M., Hawk, T.C., & Coleman, R.E. Center for the Study of Aging and Human Development, and Departments of Psychiatry and Radiology Duke University Medical Center, Durham, NC 27710 Word Length Coefficient PET Count Word Length Coefficient Striate Cortex Inferior Temporal Gyrus Word Frequency – Activation Effects Word Length – Activation Effects  In the first analysis, lexical decision RTs were averaged across participants to obtain mean RTs for each of the 720 stimulus words. Word frequency and length were then regressed onto RT.  For both age groups, word frequency coefficients were negative indicating that more familiar words were identified more quickly. Word length coefficients were positive reflecting slower responses to longer length words.  As shown above in Figure 1, although word frequency coefficients were similar for older and younger adults, word length had a larger effect on older adults’ lexical decision RTs relative to younger adults’.  Word Length effects (coefficients) were reliably related to activation only in the striate cortex of older adults.  For older adults, lower activation in the striate was associated with larger effects of length on lexical decision RT. 1 -.263 0 0.10 0.20 0.30 Word Frequency Regression Coefficient Younger Older -.249.050 Word Length.133 Word Frequency Coefficient Word Frequency Coefficient Striate Cortex Inferior Temporal Gyrus Younger Adults -.10 -.60 -.50 -.40 -.30 -.20 0.10 n.s. -.60 -.50 -.40 -.30 -.20 -.10 n.s. 0.10 Z- Score 2.43 3.07 2.43 6.22 SagittalCoronal Axial R R Younger Adults Older Adults 2 Z- Score 2.43 3.07 2.43 6.22 Sagittal Coronal Axial R R Older Adults PET Count Younger Adults Older Adults