Panksepp (1998) was the first researcher to propose a theory on the evolutionary development of emotions that accounted for the role of language. In his theory, Panksepp assumed that emotions arise in subcortical areas of the brain, are linked to learning experiences in the limbic system, and interact with higher cognitive functions including language, in the cortical areas. Thus, these three levels should play a role when looking at the affective component of words here called affective semantics. The representation of affective semantics can be examined by testing explicit valence decisions, that is, whether a word is evaluated as positive or negative. Furthermore, the relationship of language and emotions can also implicitly be tested, for example, when participants make lexical decisions on stimuli that are systematically manipulated valence. Recently, the explicit and implicit modes of valence processing have been extensively examined in adults. Consequently, the corresponding behavioral and neuronal correlates of valence are already well-established. However, it remains unclear how these explicit and implicit influences develop during childhood on both a behavioral and neural level. This dissertation seeks to address this open issue and attempts to provide answers regarding how children compared to adults process the valence of words. Three empirical studies were conducted to approach the central question of whether children between the ages of six and twelve show the same behavioral phenomena and neuronal correlates of valence processing as adults.
As a basis for the studies, a word database was designed to specifically target the vocabulary of children called the Berlin Affective Word List for Children. The words were selected based on the Berlin Affective Word List, which is a German word database for adults. The adult version of the affective word list was used as a template in order to ensure comparability between child and adult ratings. Study I (Sylvester et al., 2016) examines the question of whether children between the ages of six and twelve evaluate words in the same way as adults when they rate words in terms of their valence, arousal, and imageability. The results point to similar affective semantic representations in children and adults, including a U-shaped function relating valence and arousal ratings and an inverse U-shaped function relating valence and reaction times encompassing the positivity superiority effect (Lüdtke & Jacobs, 2015), indicating that positive words lead to shortest reaction times.
Subsequently, the neuronal correlates of explicit valence processing were examined in Study II (Sylvester et al., 2021a). According to the framework of Panksepp (1998; 2005a; 2005b), neural activation was expected in the affective semantic network which includes the amygdala, striatum, thalamus, insula, orbitofrontal, supplementary and cingulate cortex, inferior and middle frontal, superior and middle temporal gyrus. In order to test the affective semantic network, children between the ages of six and nine and adults between ages 19 and 30 performed a valence decision task. Based on the behavioral results of Study I, similar rating behavior was observed in children and adults. Furthermore, similar neural activation patterns were expected across age groups.
The neural response patterns of the children and adults of Study II are very similar to the results previously reported in adults. Specifically, we observed extensive activation within the presumed affective semantic neural network over all three valence categories (positive, negative, and neutral). However, there were differences regarding the regions predominantly recruited for valence decisions, for example there was an increased activation in amygdala in children. Moreover, activation beyond the assumed affective semantic network was observed, for example, in occipito-parietal regions, especially among adults. The different results between children and adults suggest that differences in the size of the mental lexicon between children and adults and/or the experiences from which affective semantics are derived may lead to different neural activation. It is suggested that children primarily recruit information from regions primarily associated with affective processing, whereas adults recruit regions devoted to the integration of information from various resources related to affective semantics, which includes regions in the supplementary motor cortex, middle temporal and precentral gyrus. Thus, adults seem to rely their valence decisions more strongly on multimodal affective semantic hubs since they already have integrated affective semantic information, whereas children tend to recruit valence-related information directly from the ‘source’.
In Study III (Sylvester et al., 2021b), the implicit influences of valence were assessed by using lexical decisions. The question arose as to whether the similarities found for the explicit valence decisions in Study II could also be found for implicit ones. Consequently, a similar neural pattern with activations in the large-scale affective semantic network was assumed. Similar to the results in Study II, both age cohorts showed greatly overlapping activation for processing (positive, negative and neutral) words. However, when looking more precisely at differences between the valence categories, children and adults show different processing streams. Although the adults showed similar neuronal activation as adults in previous studies, i.e., activation in orbitofrontal and anterior cingulate cortex activation (e.g., Kuchinke et al., 2005), a different picture emerged in children. The children mainly showed activation in regions associated with lexico-semantic processing, i.e., supramarginal and superior parietal gyrus activation (e.g., Price, 2012). In contrast, we observed less activation in regions directly linked to valence processing. However, positive words compare more favorably between children and adults, possibly due to that positive words are learned earlier in life (Ponari et al., 2018). These results are consistent with the interpretation of Study II in that children primarily display task-specific activation, meaning the recruitment of primarily valence-specific region during explicit valence decisions occurs whereas valence plays a subordinate role during lexical decisions. These results indicate that less connected affective semantic information is processed in children than in adults. In contrast, the adults demonstrated greater activation of valence-related areas, suggesting that valence is an important implicit source of information in lexical decisions, facilitating the decision process.
In summary, the three empirical studies of the current work are among the first to show that children and adults have similar affective semantic representations. It is, however, noteworthy that some aspects of affective semantic processing are still not adult-like in children up to the age of twelve, specifically they show more task-specific activation. Furthermore, the differences in neural activation in the affective semantic network, especially the IFG (Hofmann & Jacobs, 2014) indicate deviations between the mental lexica in children and adults. As such, the neural results in the present dissertation are interpreted in the light of differences between the children’s and adults’ mental lexicons, resulting from children’s limited experience with experiential and distributional data and/or their smaller vocabulary.
Based on the central results of this dissertation, a developmental hypothesis for affective semantics is proposed. The developmental hypothesis assumes that the amount of the apperceptive mass (Kintsch, 1980), as defined here as the size of the vocabulary and the amount of available experienced and distributed data, moderates the neural processing of affective semantics. The results could contribute to the developmental computational and cognitive models to model the influence of vocabulary on and for learning affective semantics.
