Valerie Gray Hardcastle (1995) An Expanded Role for the P300: an Addendum to Klimesch. Psycoloquy: 6(23) Memory Brain (2)

Volume: 6 (next, prev) Issue: 23 (next, prev) Article: 2 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 6(23): An Expanded Role for the P300: an Addendum to Klimesch

Commentary on Klimesch on Memory-Brain

Valerie Gray Hardcastle
Department of Philosophy
Virginia Polytechnic Institute and State University
Blacksburg, Virginia 24061-0126, USA


Klimesch argues that the P300 component in an evoked potential waveform corresponds to some sort of episodic memory search (1995). However, the P300 component must index something more than that, since we can evoke differential effects using completely novel stimuli -- stimuli that subjects could not possibly be recalling or referring to from memory.


Alpha, EEG, Hippocampus, Memory, Oscillation, Thalamus, Theta.
1. In his recent article "Memory Processes Described as Brain Oscillations in the EEG-Alpha and Theta Bands" (Klimesch, 1995), Wolfgang Klimesch reasons that "if the P300 really stems from phase-locked hippocampal theta activity, the (typical) functional meaning of the P300 should be related to the encoding of contextual and the encoding of new information" (para. 49). He goes on to (briefly) present some data that supports his line of reasoning. Though I don't want to challenge the general picture that Klimesch paints of the P300 waveform, I do think that its function should not be tied so closely to mnemonic processes. In particular, a recently completed pilot study suggests that the P300 indexes updating working memory and does not activate long term memories. (Details of the experiment can be found in Hardcastle, forthcoming).

2. As in the experiments mentioned by Klimesch, ERPs were recorded during the encoding of stimuli. However, in this case, all the stimuli were completely novel. In particular, subjects were asked to make rapid decisions concerning the physical structure of simple 5-line visual patterns in an unmasked priming paradigm. These patterns fell into three categories: closed, in which some part of the patterns formed an enclosed area; continuous, in which no portion of the patterns formed a closed area and in which the lines formed a continuous shape; and hatched, in which patterns were neither closed nor continuous. Only closed and hatched patterns were used in these experiments in order to maintain consistent shape complexity.

3. Using a reaction time paradigm, subjects were shown sets of two patterns, one following the other. They were instructed to look at, but not react to, the first shape (the prime), and then indicate whether the second shape (the target) was closed or hatched. None of the shapes were repeated for any subject. There were four types of priming relations to targets. The prime was either identical to the target, or it was rotated 180 degrees on the vertical axis, forming the "mirror" image of the target, or it fell into the same category as the target (either closed or hatched) but was not any rotation of the target shape, or it was not in the same category as the target (the unprimed condition). As expected, subjects were significantly faster to decide on the character of the shape if that shape were preceded by the same shape, a mirror image of the shape, or a different shape in the same category. Early waveform effects are what one might expect as well.

4. For my purposes here, I shall focus only on the measured P300 effects under the different priming conditions. Interestingly, all three priming conditions showed significant effects at this component. Moreover, these effects were not the same across the various priming variables. In the identity primed condition, we found that the primed P300 peaked approximately 120 msec earlier, and its amplitude was larger frontally and smaller posteriorly than the amplitude of the unprimed waveform. In contrast, the amplitude of the P300 under the mirror priming condition was less than the amplitude of the unprimed component at all electrode sites. And, surprisingly, there were no latency effects in that condition. Finally, there was also an amplitude effect with category priming. Indeed, the mean amplitude for the analyzed epoch was significantly less with category priming than with no priming. There was also a significant effect for latency, with the primed P300 peaking earlier than the unprimed one in the bilateral posterior regions.

5. These sorts of P300 modulations in an unmasked priming study are not unusual. Even though attenuation of the P300 waveform has not been found in most picture or word priming studies (see Rugg and Doyle, 1992, for review), the effect seen here is consistent with other P300 latency shifts. However, it is important to differentiate the attenuation seen in the present study and the P300 effects associated with long term memory search that Klimesch emphasizes since here subjects performed NO memory search. Indeed, they had no memories of the stimuli to search, even inadvertently. Consequently, we should not think of the P300 solely as documenting some mnemonic access. Instead, perhaps, we should conceive the P300 latency as proportional to the time required to evaluate or categorize an evoking stimulus. It seems reasonable to assume that it is easier to evaluate a target if it has been preceded by itself (or some related version of itself) (Squire, 1992),so we would expect to find a latency shift when comparing primed with unprimed waveforms.

6. Nevertheless, even though episodic memory search was not indexed, my results are still in line with Donchin's (1981) suggestion that the P300 indexes the processes by which working memory is modified in response to environmental events. He hypothesizes that the change in amplitude is proportional to the amount of change required in updating on-line representations, which would account nicely for the P300 amplitude effects seen here. Presumably, the contents of working memory would require less up-dating if the target stimulus were primed by some event useful in performing the decision task than if the target stimulus were preceded by some shape that was not helpful. In general, the amplitude of the primed condition was in fact found to be less than the unprimed condition.

7. Moreover, my results are congenial with Donchin's interpretation and not Verleger's (1988), who wants to tie to the "closure" of expectancies. We can dismiss almost immediately the suggestion that the P300 is amplified by unexpected or low frequency events, since in the investigation being reported here, each priming condition was just as likely to occur as any other. (One might argue that unprimed targets are a low frequency event relative to primed targets tout court, but in virtue of the differential priming effects seen across the different priming conditions, it seems unlikely that subjects categorized events in that way.)

8. This experiment also tells us some things about the functional identification of the P300 with respect to priming. Although the effects seen here are in congruence with the hypothesis that the P300 indexes processing in working memory, they do not dovetail with the most recent suggestions that priming occurs at a fairly low level of sensory processing in the occipital lobes (Squire et al., 1991). Recent work by Paller and Kutas (1992) indicates that priming without recollection (the sort of priming Klimesch discusses in para. 51 and the sort of priming that also must have occurred here) shows up as a positivity between 400 and 500 msec bilaterally over the posterior regions of the brain. Here, however, we found priming beginning over different areas of the brain and at different times, depending upon the type of prime subjects were given. Moreover, occipital priming effects were conspicuously absent in the early negative components, which indicates that the early sensory processing affected by priming conditions is not located in the posterior regions of the brain. The later P300 attenuation does occur bilaterally over the occipital lobes, as well as over other sites; however, we have hypothesized that this component indexes the more cognitive aspects of stimulus processing, and so these effects would not tend to corroborate Paller and Kutas's results.

9. Though nothing I have written rules out any of Klimesch's hypotheses concerning possible functions of the P300, it is important to recognize that whatever the P300 indicates, it does more than simply index "memory codes" (para. 51). For we can see differential effects in situations in which subjects are experiencing completely novel stimuli. The P300 does MORE than indicate "increased episodic memory demands" (para. 48).


Donchin, E. (1981). Surprise!...surprise. Psychophysiologia, 18, 493-513.

Hardcastle, V.G. (forthcoming). Discovering the Moment of Consciousness? An ERP Analysis of Priming Using Novel Visual Stimuli. Philosophical Psychology.

Klimesch, W. (1995). Memory Processes Described as Brain Oscillations in the EEG-Alpha and Theta Bands. PSYCOLOQUY 95(6) memory.brain.1.klimesch.

Paller, K.A. & Kutas, M. (1992). Brain potentials during memory retrieval: Neurophysiological indications of the distinction between conscious recollection and priming. Journal of Cognitive Neuroscience, 4, 375-391.

Rugg, M.D. & Doyle, M.C. (1992). In H. Heinze, T. Munte & G.R. Mangun (Eds.), Cognitive Electrophysiology. Cambridge, Massachusetts: Mirkhauser Boston.

Squire, L.R. (1992). Declarative and nondeclarative memory: Multiple brain systems supporting learning and memory. Journal of Cognitive Neuroscience, 4, 232-243.

Squire, L.R., Ojemann, J., Miezin, F., Petersen, S.E. & Raichle, M.E. (1991). The anatomy of normal human memory. MS.

Verleger, R. (1988). Event-related potentials and cognition: A critique of the context updating hypothesis and an alternative interpretation of P3. Behavioral & Brain Science, 11, 343-427.

Volume: 6 (next, prev) Issue: 23 (next, prev) Article: 2 (next prev first) Alternate versions: ASCII Summary