4.3. Brain-Behavior Relationships
Contrary to our expectations, at the inter-individual level the amount of change of the target-elicited N250 from the first to the second experiment half did not correlate with memory performance. Since these correlations were essentially zero, we suspect that this null result is not due to a lack of power or a low signal-to-noise ratio, as indicated by the high split-half reliabilities of the target-elicited N250 and the good to excellent reliability of the face memory test results. Rather it seems that the robust increase in N250 amplitude across the experiment, although possibly reflecting increasing familiarity with the designated target face, is not related to individual differences in face recognition. One conceivable reason is the ease of recognizing the target face, leading to a ceiling effect.
The relationship of the across-session changes of the N250 to non-target faces and memory performance was significant only between N250 to low-distinctive faces and performance in the easy but not in the difficult memory task. Wilhelm et al. (2010) and Hildebrandt et al. (2010) have demonstrated in large samples that face cognition in easy and difficult tasks must be distinguished, similar to the need for such a distinction when measuring general cognitive abilities. Therefore, it is an interesting finding that also the brain-behavior relationship seems to be specific for the kind of memory task – the association with the easy task was double in effect size than with the hard task. The simplest explanation for the unverifiable association between the across-session change of the N250 and performance in the difficult memory test is the lower reliability of the accuracy than of the recognition speed test. However, since our sample size was only modest and memory task performance was measured on the manifest level, it remains important to replicate these findings with a larger sample and to investigate a set of indicators for testing these associations at the latent level.
Interestingly, the brain-behavior relations of the easy task performance were found for across-session N250-changes in ERPs to low-distinctive rather than to high-distinctive faces. Intuitively, one might expect a stronger relationship between the neural signal of building-up a memory trace of high-distinctive faces and face recognition performance than for the harder to process low-distinctive faces. However, the stronger relationship found for across-session increases in low-distinctive face-elicited N250 may become plausible when considering that the ERP task required the discrimination of the low-distinctive target face from the likewise low-distinctive non-target faces. Individuals with better face recognition ability may have solved this task by a faster build-up of a representation of these non-target faces, as manifested in a larger N250 amplitude increase across the session. In contrast, memory representations for high-distinctive non-target faces may not have been necessary to make the distinction from the non-distinctive target face.
To the best of our knowledge this is the first study that shows a brain-behavior relationship between the N250 to faces and face memory. Hence, the present results extend previous findings from the N250r (Herzmann et al., 2010; Kaltwasser et al. 2014) to a condition where the N250 amplitude evolves across multiple presentations. Since the difference of the N250 across experimental halves can also be considered a kind of repetition effect, it would be of interest, whether the difference between experimental halves in the N250 is related to the N250r. This would be of particular interest, considering the close conceptual relations of these components.