CORTICAL IMAGING OF P300 COMPONENT ELICITED BY ISOLATE WORDS USING DIFFERENT MEMORY REHEARSAL STRATEGIES

L. Ding¹, A. Goldstein², J. Lian¹, E. Donchin³, B. He¹
¹Department of Bioengineering, University of Illinois at Chicago, USA
²Department of Psychology, Bar-Ilan University, Israel
³Department of Psychology, University of Illinois at Urbana-Champaign, USA

Abstract: The present study aims to identify topographic location of the neural sources underlying the verbal memory encoding, by cortical imaging analysis of the event-related potentials recorded from 11 healthy subjects participating in word encoding tasks. Both rote and elaborative strategies were used to memorize the words. The cortical potentials corresponding to the P300 component, which was elicited by the isolate words, were inversely estimated. The principal components analysis was further applied to extract major spatial components. The present results suggest that verbal memory encoding is subserved by different brain structures. Their different levels of activation may be correlated with recall performance, and may be affected by memory strategies.

INTRODUCTION

Tremendous effort has been made to elucidate neural mechanisms that participate in the implementation of episodic memory. Previous findings suggest that different neural systems participate in the episodic memory encoding, including the prefrontal cortex and the medial temporal lobe (MTL) [1-2].

The event related potential (ERP) has excellent temporal resolution desired for memory study. Different ERP components have been correlated with subsequent memory depending on the subject’s encoding strategy, on the specific nature of the stimuli and on many other factors [3]. Particularly, the P300, which is elicited by deviant events presented within the context of the oddball paradigm [3,4], has been demonstrated to be positively correlated with the subsequent recall performance when using rote rehearsal strategy. However, the scalp ERP has limited spatial resolution due to the head volume conduction effect [5]. Several EEG-based high-resolution brain imaging techniques have been developed in the past decades [6-10]. Of noteworthy is the cortical imaging technique (CIT) [10], which has been demonstrated to be able to map neural sources associated with P300 and novelty P3 components [11]. In the present study, we utilize the CIT to estimate the cortical potentials (CPs) underlying the P300 component elicited by isolate words during episodic memory encoding tasks. Then principal component analysis (PCA) is applied to the CPs for dimensionality reduction and for extraction of major spatial components [12].

METHODS

Eleven subjects were enrolled in the study under a protocol approved by the IRB/UIUC. Each participant studied 40 lists of 15 words each for subsequent memory testing. Each list contained one isolate word with larger font randomly assigned in the 6^th-10^th position. Participants were instructed to memorize the words either by silently repeating the words (rote strategy), or by connecting the words into a story (elaborative strategy). Right after the presentation of each list, the participant’s memory for the list was tested by a pencil-and-paper free recall test. The scalp potential was recorded during the presentation of the word lists with 129-channel EGI system. ERPs were obtained by averaging trials of each condition for each subject. For each of the 4 conditions (recalled or not recalled, rote or elaborate), ERPs were grand averaged across all 11 subjects for CIT analysis.

A 3-sphere inhomogeneous head model was used in the present study [10]. A closed-surface dipole layer was constructed to equivalently represent all the enclosed brain sources. The scalp potentials F_s and the equivalent dipole moments F can be linearly related by a transfer matrix A, such that AF=F_s_,Then the cortical potential F_c can be calculated by F_c=BF=BA⁺F_s_,where A⁺ is the pseudo-inverse of A, and B is the transfer matrix relating F_c and F. Then PCA was applied to an association matrix representing the covariance between each pair of the cortical observations. The covariances are computed on the epoch of P300 component, but separated by rote and elaborative strategies.

RESULTS

Figure 1 shows the midline 3-channel ERP waves elicited by the isolates, when using rote and elaborative strategies, separated as a function of whether the words were subsequently recalled or not. Clearly, isolates elicited large P300 component at about 500 ms after stimulus onset.

Figure 1. ERP waveforms elicited by the isolates, when using rote strategy (left) and elaborative strategy (right).

Figure 2 displays the top view of scalp potential maps (SPMs) and cortical potential maps (CPMs) at the peak of P300, corresponding to (a) not-recalled/rote, (b) recalled/rote, (c) not-recalled/elaborative, and (d) recalled/elaborative.

Figure 2. SMPs (top) and CMPs (bottom) at peak of P300, for not recalled/recalled isolates, with rote/elaborative strategies.

Note that the SPMs of rote strategy have blurred and symmetric positive potential distribution over the parietal lobe, while the potential distribution slightly shifts to right or frontal part for the elaborative strategy. On the other hand, more detailed spatial patterns are revealed in the estimated CPMs with activation areas of (A) left parietal lobe (LPL), (B) left inferior prefrontal lobe (LIPL), (C) right central lobe (RCL), (D) left medial temporal lobe (LMTL), and (E) right inferior prefrontal lobe (RIPL). For rote strategy, increased activity in LMTL and decreased activity in LIPL are observed in recalled as compared to not-recalled CPMs. For elaborative strategy, the CPMs show more distributed activation than the rote strategy, with more activation in the right parietal lobe for the not-recalled and more activation in the left prefrontal lobe for the recalled isolates, respectively.

The application of PCA to the CPMs results in a set of spatial factor loadings, with the first three principal components shown in Fig. 3. These three spatial factors account for total of 89.8% and 91% variance for the rote strategy (top) and elaborative strategy (bottom), respectively.

Figure 3. First three spatial factors for rote/elaborative strategies.

By examining the first principal component, the elabrative strategy is associated with more spatial activities from right parietal lobe and LIPL, but decreased activities from LMTL and RCL.

DISCUSSION

The results of the present study demonstrate that multiple brain structures may contribute to verbal memory encoding. Specifically, our results indicate that strong activition in the left MTL and LPL is highly correlated with the subsequent memory. More distributed areas of brain activation for elaborative strategy, particularly more contribution from the right parietal lobe and left prefrontal lobe, may suggest the integration and coordination of a more complex network of neural structures. Further detailed spatio-temporal analysis is desired in the future study, to illustrate the significance of different activities and their functional relationship.

ACKNOWLEDGEMENT

This work was supported in part by NSF CAREER Award BES-9875344 and NIH-MH19554.

REFERENCES

1. J.B. Brewer, Z. Zhao, J.E. Desmond, et al. “Making memories: brain activity that predicts how well visual experience will be remembered”. Science, vol. 281, pp. 1185-1187, 1998.

2. A.D. Wagner, D.L. Schacter, M. Rotte, et al. “Building memories: remembering and forgetting of verbal experiences as predicted by brain activity”. Science, vol. 281, pp. 1188-1191, 1998.

3. D. Karis, M. Fabiani, and E. Donchin. “P300 and memory: individual differences in the Von Restorff effect”. Cog. Psychol., vol. 16, pp. 177-216, 1984.

4. E. Donchin. “Surprise? … Surprise!” Psychophysiology, vol. 18, pp. 493-513, 1981.

5. P. Nunez. Electric field of the brain. Oxford University Press, London, 1981.

6. A.M. Dale and M.I. Sereno. “Improved localization of cortical activity by combining EEG and MEG with MRI cortical surface reconstruction: a linear approach”. J. Cog. Neurosci., vol. 5, pp. 162-176, 1993.

7. A. Gevins, J. Le, N. Martin, et al. “High resolution EEG: 124-channel recording, spatial deblurring and MRI integration methods”. Electroenceph. Clin. Neurophysiol., vol. 39, pp. 337-358, 1994.

8. F. Babiloni, C. Babiloni, F. Carducci, et al. “High resolution EEG: a new model-dependent spatial deblurring method using a realistically-shaped MR-constructed subject’s head model”. Electroenceph. clin. Neurophysiol., vol. 102, pp. 69-80, 1997.

9. Y. Wang and B. He. “A computer simulation study of cortical imaging from scalp potentials”. IEEE Trans. Biomed. Eng., vol. 45, pp. 724-735, 1998.

10. B. He, Y. Wang, and D. Wu. “Estimating cortical potentials from scalp EEG’s in a realistically shaped inhomogeneous head model by means of the boundary element method”. IEEE Trans. Biomed. Eng., vol. 46, pp. 1264-1268, 1999.

11. B. He, J. Lian, K.M. Spencer, et al. “A cortical potential imaging analysis of the P300 and novelty P3 components”. Hum. Brain Mapp., vol. 12, pp. 120-130, 2001.

12. Kevin M. Spencer, Joseph Dien, and Emanuel Donchin. “Spatiotemporal analysis of the late ERP responses to deviant stimuli”. Psychophysiology, vol. 38, pp. 343-358, 2001.