Learning to see the threat: temporal dynamics of ERPs of motivated attention in fear conditioning

Abstract

Social threat detection is important in everyday life. Studies of cortical activity have shown that event-related potentials (ERPs) of motivated attention are modulated during fear conditioning. The time course of motivated attention in learning and extinction of fear is, however, still largely unknown. We aimed to study temporal dynamics of learning processes in classical fear conditioning to social cues (neutral faces) by selecting an experimental setup that produces large effects on well-studied ERP components (early posterior negativity, EPN; late positive potential, LPP; stimulus preceding negativity, SPN) and then exploring small consecutive groups of trials. EPN, LPP, and SPN markedly and quickly increased during the acquisition phase in response to the CS+ but not the CS-. These changes were visible even at high temporal resolution and vanished completely during extinction. Moreover, some evidence was found for component differences in extinction learning, with differences between CS+ and CS- extinguishing faster for late as compared to early ERP components. Results demonstrate that fear learning to social cues is a very fast and highly plastic process and conceptually different ERPs of motivated attention are sensitive to these changes at high temporal resolution, pointing to specific neurocognitive and affective processes of social fear learning.

Keywords: affective neuroscience; event-related potentials; fear conditioning; motivated attention; social threat; temporal dynamics.

Fig. 1

Schematic representation of the experimental procedure and trial configuration for the performed analyses.

Fig. 2

Planar projected topographical maps of the spatio-temporal clustering. Dots indicate sensor positions as seen from above with anterior sensors at the top of the page and left sensors on the left side. Colored areas indicate samples within a significant spatio-temporal cluster. Each map corresponds roughly to 20 ms, each row to a little more than 200 ms.

Fig. 3

Regional ERP grand average waveforms elicited by the different stimuli: CS⁺, CS⁻¹ and CS⁻². Each line represents a different phase of the fear conditioning paradigm (habituation, acquisition and extinction).

Fig. 4

Learning curve reconstruction of the EPN component by consecutive subdivision of the trials per average. The three rows of topographical maps (back view on realistic head surface model) per subdivision depict the temporal trajectory of the three conditioned stimuli (CS⁺, CS⁻¹ and CS⁻²) relative to the end of the habituation phase. Line plots indicate regional amplitudes corresponding to the spatio-temporal extent of the EPN cluster (scaled and normalized to be comparable across components). Black stars indicate the level of significance of the t-tests for CS+-specific processing within the EPN component, colored stars indicate the level of significance of the t-tests for the between component comparisons. Error bars correspond to ±1 standard error (SE).

Fig. 5

Learning curve reconstruction of the LPP component by consecutive subdivision of the trials per average. The three rows of topographical maps (back view on realistic head surface model) per subdivision depict the temporal trajectory of the three conditioned stimuli (CS⁺, CS⁻¹ and CS⁻²) relative to the end of the habituation phase. Line plots indicate regional amplitudes corresponding to the spatio-temporal extent of the LPP cluster (scaled and normalized to be comparable across components). Black stars indicate the level of significance of the t-tests for CS+-specific processing within the LPP component, colored stars indicate the level of significance of the t-tests for the between component comparisons. Error bars correspond to ±1 SE.

Fig. 6

Learning curve reconstruction of the SPN component by consecutive subdivision of the trials per average. The three rows of topographical maps per subdivision (back right view on realistic head surface model) depict the temporal trajectory of the three conditioned stimuli (CS⁺, CS⁻¹ and CS⁻²) relative to the end of the habituation phase. Line plots indicate regional amplitudes corresponding to the spatio-temporal extent of the SPN cluster (scaled and normalized to be comparable across components). Stars indicate the level of significance of the t-tests for CS+-specific processing within the SPN component. Error bars correspond to ±1 SE.

Fig. 7

Means and standard errors of online valence and US expectancy ratings for the CS⁺, CS⁻¹ and CS⁻² during the eight blocks of the experiment. Low ratings on the valence scale represent more negative stimulus valence. High ratings on the US expectancy scale represent a higher expectancy of a shock US following the CS. Stars indicate the level of significance of the t-tests for CS+-specific rating (^* <0.05, ^** <0.01, ^*** <0.001, ^**** <0.0001).