Category-Specific Semantic Memory and Event Related Potentials (ERPs) Glenn Mason-Riseborough (2/10/1998) Abstract – Previous studies have indicated that different categories of visual objects presented result in different areas of activation of the brain. We measured the event related potentials in normal subjects during recognition of visual stimuli. These visual stimuli were line drawings of living or non-living objects presented on a computer monitor. We were interested in two different dimensions of study. Firstly, the processing of passive versus active response to stimuli, where passive represented no response by the subject, and active represented a keyboard response based on the type of visual stimuli presented (living or non-living). The second dimension of study was the processing of living versus non-living stimuli. We wish to address two hypotheses in this study. Firstly, do ERPs elicited by living stimuli differ (in amplitude) from those elicited by non-living stimuli? Secondly, are there hemispheric differences in the way the brain processes living versus non-living stimuli? With respect to living versus non-living stimuli, our results were calculated at two separate EEG points (electrodes 44 and 120), which corresponded respectively to left and right hemispheres (anterior to ears). The left hemisphere electrode (44) showed a significant difference (at the 5% level) in the amplitude of living versus non- living stimuli. However the right hemisphere electrode (120) did not show any significant differences in the amplitude of living versus non-living stimuli. In addition, there was no significant hemisphere differences in processing living versus non-living stimuli between these two EEG points. Introduction Recent research (eg Ferreira et al., 1997; Silveri et al., 1997; Martin et al., 1996; Perani et al., 1995; Spitzer et al., 1995) has indicated that there may be differences in the processing of distinct types of visual information. In some cases it appears that patients have knowledge loss significantly restricted to either living or non-living objects while the other categories stay intact. For example, Ferreira et al. (1997) describes three patients who were all significantly impaired in animal naming as opposed to naming tools and actions. While it seems more rare for find patients with non-living stimuli difficulties, Silveri et al. (1997) reports a patient who had significant deficits in naming non-living items as opposed to living items. While it may be the case that differences are due to poorly controlled stimulus sets, it has been suggested that these differences may be correlated with specific neurological areas. Martin et al. (1996) used positron emission tomography (PET) to measure the changes in regional cerebral blood flow (rCBF) of normal brains, associated with identifying line drawings of animals and of tools. Their results indicated that naming animals selectively activated the left medial occipital lobe. On the other hand naming tools selectively activated a left premotor area (that is also activated by imagining hand movements) and also an area in the left middle temporal gyrus. Perani et al (1995) also used PET scans of normal brains to discover differences in living versus non-living recognition. Their results indicated that non- living recognition was primarily left hemispheric, specifically of the left dorsolateral frontal cortex. In contrast, living recognition occurred bilaterally in the inferior temporo-occipital area. Spitzer et al. (1995) documents their research in this area using functional magnetic resonance imaging (fMRI). They conclude that their evidence supports the theory that semantic information is represented locally in the cortex. The present study addressed this issue of category- specific semantic memory by measuring event related potentials (ERPs) in normal subjects who were required to view living or non-living picture stimuli. This study aimed at answering two main questions. Firstly, do ERPs elicited by living stimuli differ (in amplitude) from those elicited by non-living stimuli? Secondly, are there hemispheric differences in the way the brain processes living versus non-living stimuli? Methods Fourteen right-handed human subjects participated in this study after giving their informed consent. Of these subjects, ten were male, and all subjects’ ages ranged between 19 and 32. All subjects were enrolled in a third year undergraduate neuroscience class. The subjects were fitted with an EEG electrode net to record ERPs. These nets were Electrical Geodesic Inc. 128 channel Electrode Nets. The nets contained 128 Silver/Silver Chloride (Ag/AgCl) electrodes embedded in sponge and soaked in a conductive electrolyte solution (KCl). The EEG was digitised at 250 Hz with an analogue filter bandpass of 0.1 to 39.2 Hz. Rather than using a reference electrode, average referencing was used in which the voltage at a given electrode was calculated with reference to the average voltage of all the other electrodes. During the experiment, the subjects were positioned directly in front of a standard 14 inch colour PC monitor. The subjects were asked to observe the stimulus as it was presented on the monitor. There were two viewing conditions of the stimuli –passive and active. For the passive condition the subjects were just required to observe the stimuli without responding. For the active condition the subjects were required to respond via a key-press on a computer keyboard as to whether the stimulus was living or non-living. The stimuli consisted of 100 line drawings of common objects (eg animals, body parts, household tools) taken from Snodgrass and Vanderwart (1980). Half of the objects depicted were living and the other half non-living. Living and non-living pictures were matched-pairwise for familiarity and word frequency and 23 pairs were also matched for visual complexity. During the experiment, the EEG was recording continuously, however after the completion of the experiment the data was dissected into epochs. Each epoch refers to the data recorded for a single stimulus presentation. Each epoch consisted of 1200 ms total, which included 200 ms before the onset of the stimulus (designated by an event marker) in addition to 1000 ms after the onset of the stimulus. Thus, there was a total of 300 sample points recorded per epoch. Results Regarding the first dimension of passive versus active viewing conditions the data was collapsed across living versus non-living conditions. The mean of all 128 EEG points were averaged for each subject with respect to active and passive viewing. The peak ERP (measured at approximately 520 ms after the event marker) was measured, and the difference (active minus passive) for each subject was obtained. The mean difference and standard deviations were obtained; the mean was 16.29 and the standard deviation was 17.15. A t-test showed that there was a significant difference between active and passive viewing at a 5% level of significance. For the second set of results (living versus non-living), the data was collapsed across passive versus active conditions. To see if there was a hemisphere asymmetry, data from electrodes 44 (left hemisphere) and 120 (right hemisphere) was used for all further calculations. As with the active versus passive results, the ERP was measured at 520 ms after the stimulus onset for each subject and the difference (non-living minus living) for each subject was obtained. The mean difference and standard deviations were then obtained and further t-tests were calculated. For electrode 44 the mean was 4.04 and the standard deviation was 5.31. A t-test showed that there was a significant difference between living and non-living for electrode 44 at a 5% level of significance. For electrode 120 the mean was -0.36 and the standard deviation was 11..06. A t-test showed that there was no significant difference between living and non- living for electrode 120 at a 5% level of significance. Finally, the left (electrode 44) and right (electrode 120) differences for each subject were calculated and a mean (left minus right) and standard deviation obtained. The mean was -4.39 and the standard deviation was 10.44. Again, the t-test showed that there was no significant difference between the left and right hemispheres at a 5% level of significance. Discussion While our hypotheses did not directly address the issue of the passive versus active aspect of the results, nonetheless it is useful to discuss these results regarding the data we obtained. Our results indicated that not only was there a difference in the amplitude between passive and active viewing, but that this difference was significant (at a 5% level). In other words, with 95% certainty, the difference that was found could not be accounted for just by random fluctuations. Why should this be the case? There were two components to the task: the visual feature analysis and the task response. To recall, the active condition required an additional motor response component which was not required in the passive condition. Thus, we should expect that any significant difference in the ERP amplitudes would be due to this difference. Further analysis of the data is outside the scope of this study, however it my be useful to reexamine the data at specific electrode points or groups of points. It should be remembered that the results for this study were based on averaging the ERPs for each subject across all 128 electrodes. This meant that if there were any localised differences, they may have been obscured by the remainder of the data points. Further studies may show up these localised differences. For example, we would expect that there would be greater activation in the motor cortex, after some specified length of time for the active condition. In addition, it may be useful to examine the ERPs temporally, to observe if and when the difference in amplitude occurred after the stimulus onset. Our main emphasis in this study was to discover if there was any differences in brain activation between the living and non-living stimuli. Our first hypothesis addressed the issue of whether there were any amplitude differences between living and non-living stimuli. Our results with respect to this were mixed. The results from electrode 44 showed that there was a significant difference (at the 5% level) in amplitude between processing for living as opposed to non-living stimuli. However, our results from electrode 120 indicated that there was no significant difference (at the 5% level). This indicated that there may be hemispheric differences for the processing of living versus non- living stimuli. This was the issue that the second hypothesis addressed. However, our results showed that although there was a small difference, this difference was not statistically significant (at a 5% level). Of course, it should be realised that these results only show that there is no evidence to support the hypothesis that there is a difference between living and non-living stimuli between hemispheres. Our results do not show that there is no difference. These may be contrasted with previous results (eg Perani et al., 1995) in which there was significant evidence to suggest that the left hemisphere was dominant for processing non-living stimuli, whilst living stimuli was processed bilaterally. To return to the first hypothesis, our left hemisphere (electrode 44) results are evidence that there is an amplitude difference between living and non-living stimuli. This result backs up previous studies (eg those mentioned in the introduction) that show a difference in the processing of living and non-living stimuli. However, our results with regard to the right hemisphere (electrode 120) provide no support for this theory. Thus, there is no conclusive evidence, and further data needs to be gathered to examine this issue further. It may also be useful to examine additional electrodes in the current data set. This may be achieved either by examining other individual electrodes or by averaging a number of electrodes that are spatially close. In addition, the results from the electrodes chosen may be anomalous either because of faulty electrodes or because of the specific surface location chosen. There may be other electrical interferences (eg muscles twitches). With respect to the second hypothesis, these further conditions may also be useful in providing additional information. It may be useful to compare the average ERPs from a few electrodes on the left temporal area with their counterparts on the right temporal area. References Ferreira, C. T., Guiseano, B., & Poncet, M. (1997). Category- specific anomia: Implication of different neural networks in naming. NeuroReport, 8, 1595-1602. Martin, A., Wiggs, C,. L., Ungerleider, L. G., & Haxby, J. V. (1996). Neural correlates of category-specific knowledge. Nature, 379, 649-652. Perani, D., Cappa, S. F., Bettinardi, V., Bressi, S., Gorno- Tempini, M., Matarrese, M., & Fazio, F. (1995). Different neural systems for the recognition of animals and man- made tools. NeuroReport, 6, 1637-1641. Silveri, M. C., Gainotti, G., Perani, D., Cappelletti, J. Y., Carbone, G., & Fazio, F. (1997). Naming deficit for non- living items: Neuropsychological and PET study. Neuropsychologia, 35, 3, 359-367. Snodgrass, J. G., & Vanderwart, M. (1980). A standardised set of 260 pictures: Norms for name agreement, image agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6, 174-215. Spitzer, M., Kwong, K. K., Kennedy, W., Rosen, B. R., & Belliveau, J. W. (1995). Category-specific brain activation in fMRI during picture naming. NeuroReport, 6, 2109-2112.