Visuo-spatial abstract reasoning tasks, such as Raven’s Progressive Matrices Test, have been widely used as nonverbal tests of fluid intelligence (Gf). Despite the widespread application of matrix problems and the existence... [ view full abstract ]
Visuo-spatial abstract reasoning tasks, such as Raven’s Progressive Matrices Test, have been widely used as nonverbal tests of fluid intelligence (Gf). Despite the widespread application of matrix problems and the existence of solid evidences about the spatial localization of the abstract reasoning processes involved in the solution of such a complex task, little is known about its temporal dynamics. In this study we sought to provide a finer characterization of the processes underlying visuo-spatial abstract reasoning by investigating evoked potentials (ERPs) and brain oscillatory correlates (ERSPs) of three distinct Processing Stages characterizing Gf-related problem solving, namely the (1) rule inference (RI), (2) rule application (RA) and (3) response selection (RS) stages.
Sixty young healthy participants (F=33, M=27, mean age= 23±2.4) carried out a modified version of the SANDIA matrix task (Matzen et al., 2010) during high-density electroencephalographic recording (128 channels Waveguard EEG, ANT Neuro, 2048 Hz sampling rate, 24 bit resolution). Abstract reasoning trials were presented following the three aforementioned processing stages (RI, RA, RS), with participants solving each stage in chronological order to move forward. Additionally, to test for interaction between processing stages and specific Reasoning Processes, three sets of 30 trials each were created to test distinct logic rules: conjunction (AND), disjunction (OR) and exclusive disjunction (XOR). Finally, to control for pure perceptual processing occurring irrespectively of any specific stage or process, a set of perceptual-matching problems composed by the same geometric stimuli used for AND-OR-XOR trials were included. EEG data were analyzed using EEGLAB toolbox for Matlab and ad-hoc in house scripts.
Results showed higher accuracy and overall faster reaction times for disjunction rules (OR) (0.95±0.05) in respect to conjunction AND (ACC= 0.92±0.01) and exclusive disjunction (XOR) (ACC= 0.87±0.02) rules during RI and RA stages. ERPs analysis highlighted that RI is characterized by larger P200 amplitude on the left-central region compared to RA (F1, peak amplitude: RI =1.47µV, RA=0.73µV). ERSPs analysis showed an early (~100ms from the stimulus onset) and widespread burst of theta power followed by alpha desynchronization (~300ms from stimulus onset) characterizing all the processing stages. Step-specific oscillatory patterns were also observed: (i) a mostly left-lateralized, sustained delta and theta synchronization in fronto-temporal electrodes (FT7-FT8, F3/5, F4/6) characterizes RI; a long-lasting stimulus-related desynchronization in upper-alpha power (~12 Hz) was evident during RI and RA stages in fronto-parietal leads.
In conclusion, the present data highlight stage-specific ERPs and ERSPs signatures for rule inference and rule application, thus expanding existing literature on the electrophysiological correlates of abstract reasoning. Larger frontal P200 in RI compared to RA might suggest a more significant contribution of executive abilities when inferring rather than applying a given logical rule. Moreover, results support general observation about the role of activity in the alpha, delta and theta frequency bands during abstract reasoning. In particular, results highlight the primary role played by (left) prefrontal cortex delta/theta activity during the comprehension of a given rule, as well as the importance of a widespread alpha desynchronization through the entire abstract reasoning process (RI, RA). Information about stage-specific EEG correlates allows to move forward in the identification of source of variance in individual Gf levels, as well as to possibly define more specific targets for brain electrical oscillatory stimulation and cognitive training interventions.
