Refine
Year of publication
Document Type
- Article (15)
- Doctoral Thesis (7)
- Postprint (5)
- Review (2)
- Monograph/Edited Volume (1)
Language
- English (30) (remove)
Is part of the Bibliography
- yes (30)
Keywords
- EEG (30) (remove)
Institute
- Department Psychologie (15)
- Department Linguistik (6)
- Institut für Informatik und Computational Science (3)
- Strukturbereich Kognitionswissenschaften (2)
- Department Sport- und Gesundheitswissenschaften (1)
- Extern (1)
- Fakultät für Gesundheitswissenschaften (1)
- Institut für Physik und Astronomie (1)
- Mathematisch-Naturwissenschaftliche Fakultät (1)
Visual search paradigms have provided evidence for the enhanced capture of attention by threatening faces. Especially in social anxiety, hypervigilance for threatening faces has been found repeatedly across behavioral paradigms, whose reliability however have been questioned recently. In this EEG study, we sought to determine whether the detection of threat (angry faces) is specifically enhanced in individuals with high (HSA) compared to low social anxiety (LSA). In a visual search paradigm, the N2pc component of the event-related brain potential was measured as an electrophysiological indicator of attentional selection. Twenty-one HSA and twenty-one LSA participants were investigated while searching for threatening or friendly targets within an array of neutral faces, or neutral targets within threatening or friendly distractors. Whereas no differences were found in reaction times, HSA showed significant higher detection rates for angry faces, whereas LSA showed a clear ‘happiness bias’. HSA also showed enhanced N2pc amplitudes in response to emotional facial expressions (angry and happy), indicating a general attentional bias for emotional faces. Overall, the results show that social anxiety may be characterized not only by a spatial attentional bias for threatening faces, but for emotional faces in general. In addition, the results further demonstrate the utility of the N2pc component in capturing subtle attentional biases.
The temporal dynamics of climate processes are spread across different timescales and, as such, the study of these processes at only one selected timescale might not reveal the complete mechanisms and interactions within and between the (sub-) processes. To capture the non-linear interactions between climatic events, the method of event synchronization has found increasing attention recently. The main drawback with the present estimation of event synchronization is its restriction to analysing the time series at one reference timescale only. The study of event synchronization at multiple scales would be of great interest to comprehend the dynamics of the investigated climate processes. In this paper, the wavelet-based multi-scale event synchronization (MSES) method is proposed by combining the wavelet transform and event synchronization. Wavelets are used extensively to comprehend multi-scale processes and the dynamics of processes across various timescales. The proposed method allows the study of spatio-temporal patterns across different timescales. The method is tested on synthetic and real-world time series in order to check its replicability and applicability. The results indicate that MSES is able to capture relationships that exist between processes at different timescales.
Moving Beyond ERP Components
(2018)
Relationships between neuroimaging measures and behavior provide important clues about brain function and cognition in healthy and clinical populations. While electroencephalography (EEG) provides a portable, low cost measure of brain dynamics, it has been somewhat underrepresented in the emerging field of model-based inference. We seek to address this gap in this article by highlighting the utility of linking EEG and behavior, with an emphasis on approaches for EEG analysis that move beyond focusing on peaks or "components" derived from averaging EEG responses across trials and subjects (generating the event-related potential, ERP). First, we review methods for deriving features from EEG in order to enhance the signal within single-trials. These methods include filtering based on user-defined features (i.e., frequency decomposition, time-frequency decomposition), filtering based on data-driven properties (i.e., blind source separation, BSS), and generating more abstract representations of data (e.g., using deep learning). We then review cognitive models which extract latent variables from experimental tasks, including the drift diffusion model (DDM) and reinforcement learning (RL) approaches. Next, we discuss ways to access associations among these measures, including statistical models, data-driven joint models and cognitive joint modeling using hierarchical Bayesian models (HBMs). We think that these methodological tools are likely to contribute to theoretical advancements, and will help inform our understandings of brain dynamics that contribute to moment-to-moment cognitive function.
Moving beyond ERP components
(2018)
Relationships between neuroimaging measures and behavior provide important clues about brain function and cognition in healthy and clinical populations. While electroencephalography (EEG) provides a portable, low cost measure of brain dynamics, it has been somewhat underrepresented in the emerging field of model-based inference. We seek to address this gap in this article by highlighting the utility of linking EEG and behavior, with an emphasis on approaches for EEG analysis that move beyond focusing on peaks or "components" derived from averaging EEG responses across trials and subjects (generating the event-related potential, ERP). First, we review methods for deriving features from EEG in order to enhance the signal within single-trials. These methods include filtering based on user-defined features (i.e., frequency decomposition, time-frequency decomposition), filtering based on data-driven properties (i.e., blind source separation, BSS), and generating more abstract representations of data (e.g., using deep learning). We then review cognitive models which extract latent variables from experimental tasks, including the drift diffusion model (DDM) and reinforcement learning (RL) approaches. Next, we discuss ways to access associations among these measures, including statistical models, data-driven joint models and cognitive joint modeling using hierarchical Bayesian models (HBMs). We think that these methodological tools are likely to contribute to theoretical advancements, and will help inform our understandings of brain dynamics that contribute to moment-to-moment cognitive function.
In this thesis sentence processing was investigated using a psychophysiological measure known as pupillometry as well as Event-Related Potentials (ERP). The scope of the the- sis was broad, investigating the processing of several different movement constructions with native speakers of English and second language learners of English, as well as word order and case marking in German speaking adults and children. Pupillometry and ERP allowed us to test competing linguistic theories and use novel methodologies to investigate the processing of word order. In doing so we also aimed to establish pupillometry as an effective way to investigate the processing of word order thus broadening the methodological spectrum.
The goal of a Brain-Computer Interface (BCI) consists of the development of a unidirectional interface between a human and a computer to allow control of a device only via brain signals. While the BCI systems of almost all other groups require the user to be trained over several weeks or even months, the group of Prof. Dr. Klaus-Robert Müller in Berlin and Potsdam, which I belong to, was one of the first research groups in this field which used machine learning techniques on a large scale. The adaptivity of the processing system to the individual brain patterns of the subject confers huge advantages for the user. Thus BCI research is considered a hot topic in machine learning and computer science. It requires interdisciplinary cooperation between disparate fields such as neuroscience, since only by combining machine learning and signal processing techniques based on neurophysiological knowledge will the largest progress be made. In this work I particularly deal with my part of this project, which lies mainly in the area of computer science. I have considered the following three main points: <b>Establishing a performance measure based on information theory:</b> I have critically illuminated the assumptions of Shannon's information transfer rate for application in a BCI context. By establishing suitable coding strategies I was able to show that this theoretical measure approximates quite well to what is practically achieveable. <b>Transfer and development of suitable signal processing and machine learning techniques:</b> One substantial component of my work was to develop several machine learning and signal processing algorithms to improve the efficiency of a BCI. Based on the neurophysiological knowledge that several independent EEG features can be observed for some mental states, I have developed a method for combining different and maybe independent features which improved performance. In some cases the performance of the combination algorithm outperforms the best single performance by more than 50 %. Furthermore, I have theoretically and practically addressed via the development of suitable algorithms the question of the optimal number of classes which should be used for a BCI. It transpired that with BCI performances reported so far, three or four different mental states are optimal. For another extension I have combined ideas from signal processing with those of machine learning since a high gain can be achieved if the temporal filtering, i.e., the choice of frequency bands, is automatically adapted to each subject individually. <b>Implementation of the Berlin brain computer interface and realization of suitable experiments:</b> Finally a further substantial component of my work was to realize an online BCI system which includes the developed methods, but is also flexible enough to allow the simple realization of new algorithms and ideas. So far, bitrates of up to 40 bits per minute have been achieved with this system by absolutely untrained users which, compared to results of other groups, is highly successful.
Intuitively, it is clear that neural processes and eye movements in reading are closely connected, but only few studies have investigated both signals simultaneously. Instead, the usual approach is to record them in separate experiments and to subsequently consolidate the results. However, studies using this approach have shown that it is feasible to coregister eye movements and EEG in natural reading and contributed greatly to the understanding of oculomotor processes in reading. The present thesis builds upon that work, assessing to what extent coregistration can be helpful for sentence processing research.
In the first study, we explore how well coregistration is suited to study subtle effects common to psycholinguistic experiments by investigating the effect of distance on dependency resolution. The results demonstrate that researchers must improve the signal-to-noise ratio to uncover more subdued effects in coregistration. In the second study, we compare oscillatory responses in different presentation modes. Using robust effects from world knowledge violations, we show that the generation and retrieval of memory traces may differ between natural reading and word-by-word presentation. In the third study, we bridge the gap between our knowledge of behavioral and neural responses to integration difficulties in reading by analyzing the EEG in the context of regressive saccades. We find the P600, a neural indicator of recovery processes, when readers make a regressive saccade in response to integration difficulties.
The results in the present thesis demonstrate that coregistration can be a useful tool for the study of sentence processing. However, they also show that it may not be suitable for some questions, especially if they involve subtle effects.
Background: Empirical evidence suggests substantial deficits regarding emotion recognition in bulimia nervosa (BN). The aim of the current study was to investigate electrophysiologic evidence for deficits in emotional face processing in patients with BN. Methods: Event-related potentials were recorded from 13 women with BN and 13 matched healthy controls while viewing neutral, happy, fearful, and angry facial expressions. Participants' recognition performance for emotional faces was tested in a subsequent categorization task. In addition, the degree of alexithymia, depression, and anxiety were assessed via questionnaires. Results: Categorization of emotional faces was hampered in BN (p = .01). Amplitudes of event-related potentials differed during emotional face processing: face-specific N170 amplitudes were less pronounced for angry faces in patients with BN (mean [M] [standard deviation {SD}] = 1.46 [0.56] mu V versus M [SD] = -1.23 [0.61] mu V, p = .02). In contrast, P3 amplitudes were more pronounced in patients with BN as compared with controls (M [SD] = 2.64 [0.46] mu V versus M [SD] = 1.25 [0.39] mu V, p = .04), independent of emotional expression. Conclusions: The study provides novel electrophysiologic data showing that emotional faces are processed differently in patients with BN as compared with healthy controls. We suggest that deficits in early automatic emotion classification in BN are followed by an increased allocation of attentional resources to compensate for those deficits. These findings might contribute to a better understanding of the impaired social functioning in BN.
Behavioral research has shown that infants use both behavioral cues and verbal cues when processing the goals of others' actions. For instance, 18-month-olds selectively imitate an observed goal-directed action depending on its (in)congruence with a model's previous verbal announcement of a desired action goal. This EEG-study analyzed the electrophysiological underpinnings of these behavioral findings on the two functional levels of conceptual action processing and motor activation. Mid-latency mean negative ERP amplitude and mu-frequency band power were analyzed while 18-month-olds (N = 38) watched videos of an adult who performed one out of two potential actions on a novel object. In a within-subjects design, the action demonstration was preceded by either a congruent or an incongruent verbally announced action goal (e.g., "up" or "down" and upward movement). Overall, ERP negativity did not differ between conditions, but a closer inspection revealed that in two subgroups, about half of the infants showed a broadly distributed increased mid-latency ERP negativity (indicating enhanced conceptual action processing) for either the congruent or the incongruent stimuli, respectively. As expected, mu power at sensorimotor sites was reduced (indicating enhanced motor activation) for congruent relative to incongruent stimuli in the entire sample. Both EEG correlates were related to infants' language skills. Hence, 18-month-olds integrate action-goal-related verbal cues into their processing of others' actions, at the functional levels of both conceptual processing and motor activation. Further, cue integration when inferring others' action goals is related to infants' language proficiency.
Recent research suggests that the P3b may be closely related to the activation of the locus coeruleus-norepinephrine (LC-NE) system. To further study the potential association, we applied a novel technique, the non-invasive transcutaneous vagus nerve stimulation (tVNS), which is speculated to increase noradrenaline levels. Using a within-subject cross-over design, 20 healthy participants received continuous tVNS and sham stimulation on two consecutive days (stimulation counterbalanced across participants) while performing a visual oddball task. During stimulation, oval non-targets (standard), normal-head (easy) and rotated-head (difficult) targets, as well as novel stimuli (scenes) were presented. As an indirect marker of noradrenergic activation we also collected salivary alpha-amylase (sAA) before and after stimulation. Results showed larger P3b amplitudes for target, relative to standard stimuli, irrespective of stimulation condition. Exploratory post hoc analyses, however, revealed that, in comparison to standard stimuli, easy (but not difficult) targets produced larger P3b (but not P3a) amplitudes during active tVNS, compared to sham stimulation. For sAA levels, although main analyses did not show differential effects of stimulation, direct testing revealed that tVNS (but not sham stimulation) increased sAA levels after stimulation. Additionally, larger differences between tVNS and sham stimulation in P3b magnitudes for easy targets were associated with larger increase in sAA levels after tVNS, but not after sham stimulation. Despite preliminary evidence for a modulatory influence of tVNS on the P3b, which may be partly mediated by activation of the noradrenergic system, additional research in this field is clearly warranted. Future studies need to clarify whether tVNS also facilitates other processes, such as learning and memory, and whether tVNS can be used as therapeutic tool.
Recent research suggests that the P3b may be closely related to the activation of the locus coeruleus-norepinephrine (LC-NE) system. To further study the potential association, we applied a novel technique, the non-invasive transcutaneous vagus nerve stimulation (tVNS), which is speculated to increase noradrenaline levels. Using a within-subject cross-over design, 20 healthy participants received continuous tVNS and sham stimulation on two consecutive days (stimulation counterbalanced across participants) while performing a visual oddball task. During stimulation, oval non-targets (standard), normal-head (easy) and rotated-head (difficult) targets, as well as novel stimuli (scenes) were presented. As an indirect marker of noradrenergic activation we also collected salivary alpha-amylase (sAA) before and after stimulation. Results showed larger P3b amplitudes for target, relative to standard stimuli, irrespective of stimulation condition. Exploratory post hoc analyses, however, revealed that, in comparison to standard stimuli, easy (but not difficult) targets produced larger P3b (but not P3a) amplitudes during active tVNS, compared to sham stimulation. For sAA levels, although main analyses did not show differential effects of stimulation, direct testing revealed that tVNS (but not sham stimulation) increased sAA levels after stimulation. Additionally, larger differences between tVNS and sham stimulation in P3b magnitudes for easy targets were associated with larger increase in sAA levels after tVNS, but not after sham stimulation. Despite preliminary evidence for a modulatory influence of tVNS on the P3b, which may be partly mediated by activation of the noradrenergic system, additional research in this field is clearly warranted. Future studies need to clarify whether tVNS also facilitates other processes, such as learning and memory, and whether tVNS can be used as therapeutic tool.
Physical fatigue (PF) negatively affects postural control, resulting in impaired balance performance in young and older adults. Similar effects on postural control can be observed for mental fatigue (MF) mainly in older adults. Controversial results exist for young adults. There is a void in the literature on the effects of fatigue on balance and cortical activity. Therefore, this study aimed to examine the acute effects of PF and MF on postural sway and cortical activity. Fifteen healthy young adults aged 28 ± 3 years participated in this study. MF and PF protocols comprising of an all-out repeated sit-to-stand task and a computer-based attention network test, respectively, were applied in random order. Pre and post fatigue, cortical activity and postural sway (i.e., center of pressure displacements [CoPd], velocity [CoPv], and CoP variability [CV CoPd, CV CoPv]) were tested during a challenging bipedal balance board task. Absolute spectral power was calculated for theta (4–7.5 Hz), alpha-2 (10.5–12.5 Hz), beta-1 (13–18 Hz), and beta-2 (18.5–25 Hz) in frontal, central, and parietal regions of interest (ROI) and baseline-normalized. Inference statistics revealed a significant time-by-fatigue interaction for CoPd (p = 0.009, d = 0.39, Δ 9.2%) and CoPv (p = 0.009, d = 0.36, Δ 9.2%), and a significant main effect of time for CoP variability (CV CoPd: p = 0.001, d = 0.84; CV CoPv: p = 0.05, d = 0.62). Post hoc analyses showed a significant increase in CoPd (p = 0.002, d = 1.03) and CoPv (p = 0.003, d = 1.03) following PF but not MF. For cortical activity, a significant time-by-fatigue interaction was found for relative alpha-2 power in parietal (p < 0.001, d = 0.06) areas. Post hoc tests indicated larger alpha-2 power increases after PF (p < 0.001, d = 1.69, Δ 3.9%) compared to MF (p = 0.001, d = 1.03, Δ 2.5%). In addition, changes in parietal alpha-2 power and measures of postural sway did not correlate significantly, irrespective of the applied fatigue protocol. No significant changes were found for the other frequency bands, irrespective of the fatigue protocol and ROI under investigation. Thus, the applied PF protocol resulted in increased postural sway (CoPd and CoPv) and CoP variability accompanied by enhanced alpha-2 power in the parietal ROI while MF led to increased CoP variability and alpha-2 power in our sample of young adults. Potential underlying cortical mechanisms responsible for the greater increase in parietal alpha-2 power after PF were discussed but could not be clearly identified as cause. Therefore, further future research is needed to decipher alternative interpretations.
Physical fatigue (PF) negatively affects postural control, resulting in impaired balance performance in young and older adults. Similar effects on postural control can be observed for mental fatigue (MF) mainly in older adults. Controversial results exist for young adults. There is a void in the literature on the effects of fatigue on balance and cortical activity. Therefore, this study aimed to examine the acute effects of PF and MF on postural sway and cortical activity. Fifteen healthy young adults aged 28 ± 3 years participated in this study. MF and PF protocols comprising of an all-out repeated sit-to-stand task and a computer-based attention network test, respectively, were applied in random order. Pre and post fatigue, cortical activity and postural sway (i.e., center of pressure displacements [CoPd], velocity [CoPv], and CoP variability [CV CoPd, CV CoPv]) were tested during a challenging bipedal balance board task. Absolute spectral power was calculated for theta (4–7.5 Hz), alpha-2 (10.5–12.5 Hz), beta-1 (13–18 Hz), and beta-2 (18.5–25 Hz) in frontal, central, and parietal regions of interest (ROI) and baseline-normalized. Inference statistics revealed a significant time-by-fatigue interaction for CoPd (p = 0.009, d = 0.39, Δ 9.2%) and CoPv (p = 0.009, d = 0.36, Δ 9.2%), and a significant main effect of time for CoP variability (CV CoPd: p = 0.001, d = 0.84; CV CoPv: p = 0.05, d = 0.62). Post hoc analyses showed a significant increase in CoPd (p = 0.002, d = 1.03) and CoPv (p = 0.003, d = 1.03) following PF but not MF. For cortical activity, a significant time-by-fatigue interaction was found for relative alpha-2 power in parietal (p < 0.001, d = 0.06) areas. Post hoc tests indicated larger alpha-2 power increases after PF (p < 0.001, d = 1.69, Δ 3.9%) compared to MF (p = 0.001, d = 1.03, Δ 2.5%). In addition, changes in parietal alpha-2 power and measures of postural sway did not correlate significantly, irrespective of the applied fatigue protocol. No significant changes were found for the other frequency bands, irrespective of the fatigue protocol and ROI under investigation. Thus, the applied PF protocol resulted in increased postural sway (CoPd and CoPv) and CoP variability accompanied by enhanced alpha-2 power in the parietal ROI while MF led to increased CoP variability and alpha-2 power in our sample of young adults. Potential underlying cortical mechanisms responsible for the greater increase in parietal alpha-2 power after PF were discussed but could not be clearly identified as cause. Therefore, further future research is needed to decipher alternative interpretations.
Cognitive or executive control is a critical mental ability, an important marker of mental illness, and among the most heritable of neurocognitive traits. Two candidate genes, catechol-O-methyltransferase (COMT) and DRD4, which both have a roles in the regulation of cortical dopamine, have been consistently associated with cognitive control. Here, we predicted that individuals with the COMT Met/Met allele would show improved response execution and inhibition as indexed by event-related potentials in a Go/NoGo task, while individuals with the DRD4 7-repeat allele would show impaired brain activity. We used independent component analysis (ICA) to separate brain source processes contributing to high-density EEG scalp signals recorded during the task. As expected, individuals with the DRD4 7-repeat polymorphism had reduced parietal P3 source and scalp responses to response (Go) compared to those without the 7-repeat. Contrary to our expectation, the COMT homozygous Met allele was associated with a smaller frontal P3 source and scalp response to response-inhibition (NoGo) stimuli, suggesting that while more dopamine in frontal cortical areas has advantages in some tasks, it may also compromise response inhibition function. An interaction effect emerged for P3 source responses to Go stimuli. These were reduced in those with both the 7-repeat DRD4 allele and either the COMT Val/Val or the Met/Met homozygous polymorphisms but not in those with the heterozygous Val/Met polymorphism. This epistatic interaction between DRD4 and COMT replicates findings that too little or too much dopamine impairs cognitive control. The anatomic and functional separated maximally independent cortical EEG sources proved more informative than scalp channel measures for genetic studies of brain function and thus better elucidate the complex mechanisms in psychiatric illness.
Brain-electric correlates of reading have traditionally been studied with word-by-word presentation, a condition that eliminates important aspects of the normal reading process and precludes direct comparisons between neural activity and oculomotor behavior. In the present study, we investigated effects of word predictability on eye movements (EM) and fixation-related brain potentials (FRPs) during natural sentence reading. Electroencephalogram (EEG) and EM (via video-based eye tracking) were recorded simultaneously while subjects read heterogeneous German sentences, moving their eyes freely over the text. FRPs were time-locked to first-pass reading fixations and analyzed according to the cloze probability of the currently fixated word. We replicated robust effects of word predictability on EMs and the N400 component in FRPs. The data were then used to model the relation among fixation duration, gaze duration, and N400 amplitude, and to trace the time course of EEG effects relative to effects in EM behavior. In an extended Methodological Discussion section, we review 4 technical and data-analytical problems that need to be addressed when FRPs are recorded in free-viewing situations (such as reading, visual search, or scene perception) and propose solutions. Results suggest that EEG recordings during normal vision are feasible and useful to consolidate findings from EEG and eye-tracking studies.
Combined optimization of spatial and temporal filters for improving brain-computer interfacing
(2006)
Brain-computer interface (BCI) systems create a novel communication channel from the brain to an output de ice by bypassing conventional motor output pathways of nerves and muscles. Therefore they could provide a new communication and control option for paralyzed patients. Modern BCI technology is essentially based on techniques for the classification of single-trial brain signals. Here we present a novel technique that allows the simultaneous optimization of a spatial and a spectral filter enhancing discriminability rates of multichannel EEG single-trials. The evaluation of 60 experiments involving 22 different subjects demonstrates the significant superiority of the proposed algorithm over to its classical counterpart: the median classification error rate was decreased by 11%. Apart from the enhanced classification, the spatial and/or the spectral filter that are determined by the algorithm can also be used for further analysis of the data, e.g., for source localization of the respective brain rhythms.
In reading, word frequency is commonly regarded as the major bottom-up determinant for the speed of lexical access. Moreover, language processing depends on top-down information, such as the predictability of a word from a previous context. Yet, however, the exact role of top-down predictions in visual word recognition is poorly understood: They may rapidly affect lexical processes, or alternatively, influence only late post-lexical stages. To add evidence about the nature of top-down processes and their relation to bottom-up information in the timeline of word recognition, we examined influences of frequency and predictability on event-related potentials (ERPs) in several sentence reading studies. The results were related to eye movements from natural reading as well as to models of word recognition. As a first and major finding, interactions of frequency and predictability on ERP amplitudes consistently revealed top-down influences on lexical levels of word processing (Chapters 2 and 4). Second, frequency and predictability mediated relations between N400 amplitudes and fixation durations, pointing to their sensitivity to a common stage of word recognition; further, larger N400 amplitudes entailed longer fixation durations on the next word, a result providing evidence for ongoing processing beyond a fixation (Chapter 3). Third, influences of presentation rate on ERP frequency and predictability effects demonstrated that the time available for word processing critically co-determines the course of bottom-up and top-down influences (Chapter 4). Fourth, at a near-normal reading speed, an early predictability effect suggested the rapid comparison of top-down hypotheses with the actual visual input (Chapter 5). The present results are compatible with interactive models of word recognition assuming that early lexical processes depend on the concerted impact of bottom-up and top-down information. We offered a framework that reconciles the findings on a timeline of word recognition taking into account influences of frequency, predictability, and presentation rate (Chapter 4).
This thesis is concerned with the solution of the blind source separation problem (BSS). The BSS problem occurs frequently in various scientific and technical applications. In essence, it consists in separating meaningful underlying components out of a mixture of a multitude of superimposed signals. In the recent research literature there are two related approaches to the BSS problem: The first is known as Independent Component Analysis (ICA), where the goal is to transform the data such that the components become as independent as possible. The second is based on the notion of diagonality of certain characteristic matrices derived from the data. Here the goal is to transform the matrices such that they become as diagonal as possible. In this thesis we study the latter method of approximate joint diagonalization (AJD) to achieve a solution of the BSS problem. After an introduction to the general setting, the thesis provides an overview on particular choices for the set of target matrices that can be used for BSS by joint diagonalization. As the main contribution of the thesis, new algorithms for approximate joint diagonalization of several matrices with non-orthogonal transformations are developed. These newly developed algorithms will be tested on synthetic benchmark datasets and compared to other previous diagonalization algorithms. Applications of the BSS methods to biomedical signal processing are discussed and exemplified with real-life data sets of multi-channel biomagnetic recordings.
Electroencephalographic (EEG) research indicates changes in adults' low frequency bands of frontoparietal brain areas executing different balance tasks with increasing postural demands. However, this issue is unsolved for adolescents when performing the same balance task with increasing difficulty. Therefore, we examined the effects of a progressively increasing balance task difficulty on balance performance and brain activity in adolescents. Thirteen healthy adolescents aged 16-17 year performed tests in bipedal upright stance on a balance board with six progressively increasing levels of task difficulty. Postural sway and cortical activity were recorded simultaneously using a pressure sensitive measuring system and EEG. The power spectrum was analyzed for theta (4-7 Hz) and alpha-2 (10-12 Hz) frequency bands in pre-defined frontal, central, and parietal clusters of electrocortical sources. Repeated measures analysis of variance (rmANOVA) showed a significant main effect of task difficulty for postural sway (p < 0.001; d = 6.36). Concomitantly, the power spectrum changed in frontal, bilateral central, and bilateral parietal clusters. RmANOVAs revealed significant main effects of task difficulty for theta band power in the frontal (p < 0.001, d = 1.80) and both central clusters (left: p < 0.001, d = 1.49; right: p < 0.001, d = 1.42) as well as for alpha-2 band power in both parietal clusters (left: p < 0.001, d = 1.39; right: p < 0.001, d = 1.05) and in the central right cluster (p = 0.005, d = 0.92). Increases in theta band power (frontal, central) and decreases in alpha-2 power (central, parietal) with increasing balance task difficulty may reflect increased attentional processes and/or error monitoring as well as increased sensory information processing due to increasing postural demands. In general, our findings are mostly in agreement with studies conducted in adults. Similar to adult studies, our data with adolescents indicated the involvement of frontoparietal brain areas in the regulation of postural control. In addition, we detected that activity of selected brain areas (e.g., bilateral central) changed with increasing postural demands.
The PhD thesis entitled “Actions through the lens of communicative cues. The influence of verbal cues and emotional cues on action processing and action selection in the second year of life” is based on four studies, which examined the cognitive integration of another person’s communicative cues (i.e., verbal cues, emotional cues) with behavioral cues in 18- and 24-month-olds. In the context of social learning of instrumental actions, it was investigated how the intention-related coherence of either a verbally announced action intention or an emotionally signaled action evaluation with an action demonstration influenced infants’ neuro-cognitive processing (Study I) and selection (Studies II, III, IV) of a novel object-directed action. Developmental research has shown that infants benefit from another’s behavioral cues (e.g., action effect, persistency, selectivity) to infer the underlying goal or intention, respectively, of an observed action (e.g., Cannon & Woodward, 2012; Woodward, 1998). Particularly action effects support infants in distinguishing perceptual action features (e.g., target object identity, movement trajectory, final target object state) from conceptual action features such as goals and intentions. However, less is known about infants’ ability to cognitively integrate another’s behavioral cues with additional action-related communicative cues. There is some evidence showing that in the second year of life, infants selectively imitate a novel action that is verbally (“There!”) or emotionally (positive expression) marked as aligning with the model’s action intention over an action that is verbally (“Whoops!”) or emotionally (negative expression) marked as unintentional (Carpenter, Akhtar, & Tomasello, 1998; Olineck & Poulin-Dubois, 2005, 2009; Repacholi, 2009; Repacholi, Meltzoff, Toub, & Ruba, 2016). Yet, it is currently unclear which role the specific intention-related coherence of a communicative cue with a behavioral cue plays in infants’ action processing and action selection that is, whether the communicative cue confirms, contrasts, clarifies, or is unrelated to the behavioral cue. Notably, by using both verbal cues and emotional cues, we examined not only two domains of communicative cues but also two qualitatively distinct relations between behavioral cues on the one hand and communicative cues on the other hand. More specifically, a verbal cue has the potential to communicate an action intention in the absence of an action demonstration and thus a prior-intention (Searle, 1983), whereas an emotional cue evaluates an ongoing or past action demonstration and thus signals an intention-in-action (Searle, 1983). In a first research focus, this thesis examined infants’ capacity to cognitively integrate another’s intention-related communicative cues and behavioral cues, and also focused on the role of the social cues’ coherence in infants’ action processing and action selection. In a second research focus, and to gain more elaborate insights into how the sub-processes of social learning (attention, encoding, response; cf. Bandura, 1977) are involved in this coherence-sensitive integrative processing, we employed a multi-measures approach. More specifically, we used Electroencephalography (EEG) and looking times to examine how the cues’ coherence influenced the compound of attention and encoding, and imitation (including latencies to first-touch and first-action) to address the compound of encoding and response. Based on the action-reconstruction account (Csibra, 2007), we predicted that infants use extra-motor information (i.e., communicative cues) together with behavioral cues to reconstruct another’s action intention. Accordingly, we expected infants to possess a flexibly organized internal action hierarchy, which they adapt according to the cues’ coherence that is, according to what they inferred to be the overarching action goal. More specifically, in a social-learning situation that comprised an adult model, who demonstrated an action on a novel object that offered two actions, we expected the demonstrated action to lead infants’ action hierarchy when the communicative (i.e., verbal, emotional) cue conveyed similar (confirming coherence) or no additional (un-related coherence) intention-related information relative to the behavioral cue. In terms of action selection, this action hierarchy should become evident in a selective imitation of the demonstrated action. However, when the communicative cue questioned (contrasting coherence) the behaviorally implied action goal or was the only cue conveying meaningful intention-related information (clarifying coherence), the verbally/emotionally intended action should ascend infants’ action hierarchy. Consequently, infants’ action selection should align with the verbally/emotionally intended action (goal emulation). Notably, these predictions oppose the direct-matching perspective (Rizzolatti & Craighero, 2004), according to which the observation of another’s action directly resonates with the observer’s motor repertoire, with this motor resonance enabling the identification of the underlying action goal. Importantly, the direct-matching perspective predicts a rather inflexible action hierarchy inasmuch as the process of goal identification should solely rely on the behavioral cue, irrespective of the behavioral cue’s coherence with extra-motor intention-related information, as it may be conveyed via communicative cues. As to the role of verbal cues, Study I used EEG to examine the influence of a confirming (Congruent) versus contrasting (Incongruent) coherence of a verbal action intention with the same action demonstration on 18-month-olds’ conceptual action processing (as measured via mid-latency mean negative ERP amplitude) and motor activation (as measured via central mu-frequency band power). The action was demonstrated on a novel object that offered two action alternatives from a neutral position. We expected mid-latency ERP negativity to be enhanced in Incongruent compared to Congruent, because past EEG research has demonstrated enhanced conceptual processing for stimuli that mismatched rather than matched the semantic context (Friedrich & Friederici, 2010; Kaduk et al., 2016). Regarding motor activation, Csibra (2007) posited that the identification of a clear action goal constitutes a crucial basis for motor activation to occur. We therefore predicted reduced mu power (indicating enhanced motor activation) for Congruent than Incongruent, because in Congruent, the cues’ match provides unequivocal information about the model’s action goal, whereas in Incongruent, the conflict may render the model’s action goal more unclear. Unexpectedly, in the entire sample, 18-month-olds’ mid-latency ERP negativity during the observation of the same action demonstration did not differ significantly depending on whether this action was congruent or incongruent with the model’s verbal action intention. Yet, post hoc analyses revealed the presence of two subgroups of infants, each of which exhibited significantly different mid-latency ERP negativity for Congruent versus Incongruent, but in opposing directions. The subgroups differed in their productive action-related language skills, with the linguistically more advanced infants exhibiting the expected response pattern of enhanced ERP mean negativity in Incongruent than Congruent, indicating enhanced conceptual processing of an action demonstration that was contrasted rather than confirmed by the verbal action context. As expected, central mu power in the entire sample was reduced in Congruent relative to Incongruent, indicating enhanced motor activation when the action demonstration was preceded by a confirming relative to a contrasting verbal action intention. This finding may indicate the covert preparation for a preferential imitation of the congruent relative to the incongruent action (Filippi et al., 2016; Frey & Gerry, 2006). Overall, these findings are in line with the action-reconstruction account (Csibra, 2007), because they suggest a coherence-sensitive attention to and encoding of the same perceptual features of another’s behavior and thus a cognitive integration of intention-related verbal cues and behavioral cues. Yet, because the subgroup constellation in infants’ ERPs was only discovered post hoc, future research is clearly required to substantiate this finding. Also, future research should validate our interpretation that enhanced motor activation may reflect an electrophysiological marker of subsequent imitation by employing EEG and imitation in a within-subjects design. Study II built on Study I by investigating the impact of coherence of a verbal cue and a behavioral cue on 18- and 24-month-olds’ action selection in an imitation study. When infants of both age groups observed a confirming (Congruent) or unrelated (Pseudo-word: action demonstration was associated with novel verb-like cue) coherence, they selectively imitated the demonstrated action over the not demonstrated, alternative action, with no difference between these two conditions. These findings suggest that, as expected, infants’ action hierarchy was led by the demonstrated action when the verbal cue provided similar (Congruent) or no additional (Pseudo-word) intention-related information relative to a meaningful behavioral cue. These findings support the above-mentioned interpretation that enhanced motor activation during action observation may reflect a covert preparation for imitation (Study I). Interestingly, infants did not seem to benefit from the intention-highlighting effect of the verbal cue in Congruent, suggesting that the verbal cue had an unspecific (e.g., attention-guiding) effect on infants’ action selection. Contrary, when infants observed a contrasting (Incongruent) or clarifying (Failed-attempt: model failed to manipulate the object but verbally announced a certain action intention) coherence, their action selection varied with age and also varied across the course of the experiment (block 1 vs. block 2). More specifically, the 24-month-olds made stronger use of the verbal cue for their action selection in block 1 than did the 18-month-olds. However, while the 18-month-olds’ use of the verbal cue increased across blocks, particularly in Incongruent, the 24-month-olds’ use of the verbal cue decreased across blocks. Overall, these results suggest that, as expected, infants’ action hierarchy in Incongruent (both age groups) and Failed-attempt (only 24-month-olds) drew on the verbal action intention, because in both age groups, infants emulated the verbal intention about as often as they imitated the demonstrated action or even emulated the verbal action intention preferentially. Yet, these findings were confined to certain blocks. It may be argued that the younger age group had a harder time inferring and emulating the intended, yet never observed action, because this requirement is more demanding in cognitive and motor terms. These demands may explain why the 18-month-olds needed some time to take account of the verbal action intention. Contrary, it seems that the 24-month-olds, although demonstrating their principle capacity to take account of the verbal cue in block 1, lost trust in the model’s verbal cue, maybe because the verbal cue did not have predictive value for the model’s actual behavior. Supporting this interpretation, research on selective trust has demonstrated that already infants evaluate another’s reliability or competence, respectively, based on how that model handles familiar objects (behavioral reliability) or labels familiar objects (verbal reliability; for reviews, see Mills, 2013; Poulin-Dubois & Brosseau-Liard, 2016). Relatedly, imitation research has demonstrated that the interpersonal aspects of a social-learning situation gain increasing relevance for infants during the second year of life (Gellén & Buttelmann, 2019; Matheson, Moore, & Akhtar, 2013; Uzgiris, 1981). It may thus be argued that when the 24-month-olds were repeatedly faced with a verbally unreliable model, they de-evaluated the verbal cue as signaling the model’s action intention and instead relied more heavily on alternative cues such as the behavioral cue (Incongruent) or the action context (e.g., object affordances, salience; Failed-attempt). Infants’ first-action latencies were higher in Incongruent and Failed-attempt than in both Congruent and Pseudo-word, and were also higher in Failed-attempt than in Incongruent. These latency-findings thus indicate that situations involving a meaningful verbal cue that deviated from the behavioral cue are cognitively more demanding, resulting in a delayed initiation of a behavioral response. In sum, the findings of Study II suggest that both age groups were highly flexible in their integration of a verbal cue and behavioral cue. Moreover, our results do not indicate a general superiority of either cue. Instead, it seems to depend on the informational gain conveyed by the verbal cue whether it exerts a specific, intention-highlighting effect (Incongruent, Failed-attempt) or an unspecific (e.g., attention-guiding) effect (Congruent, Pseudo-word). Studies III and IV investigated the impact of another’s action-related emotional cues on 18-month-olds’ action selection. In Study III, infants observed a model, who demonstrated two actions on a novel object in direct succession, and who combined one of the two actions with a positive (happy) emotional expression and the other action with a negative (sad) emotional expression. As expected, infants imitated the positively emoted (PE) action more often than the negatively emoted (NE) action. This preference arose from an increase in infants’ readiness to perform the PE action from the baseline period (prior to the action demonstrations) to the test period (following the action demonstrations), rather than from a decrease in readiness to the perform the NE action. The positive cue thus had a stronger behavior-regulating effect than the negative cue. Notably, infants’ more general object-directed behavior in terms of first-touch latencies remained unaffected by the emotional cues’ valence, indicating that infants had linked the emotional cues specifically to the corresponding action and not the object as a whole (Repacholi, 2009). Also, infants’ looking times during the action demonstration did not differ significantly as a function of emotional valence and were characterized by a predominant attentional focus to the action/object rather than to the model’s face. Together with the findings on infants’ first-touch latencies, these results indicate a sensitivity for the notion that emotions can have very specific referents (referential specificity; Martin, Maza, McGrath, & Phelps, 2014). Together, Study III provided evidence for selective imitation based on another’s intention-related (particularly positive) emotional cues in an action-selection task, and thus indicates that infants’ action hierarchy flexibly responds to another’s emotional evaluation of observed actions. According to Repacholi (2009), we suggest that infants used the model’s emotional evaluation to re-appraise the corresponding action (effect), for instance in terms of desirability. Study IV followed up on Study III by investigating the role of the negative emotional cue for infants’ action selection in more detail. Specifically, we investigated whether a contrasting (negative) emotional cue alone would be sufficient to differentially rank the two actions along infants’ action hierarchy or whether instead infants require direct information about the model’s action intention (in the form of a confirming action-emotion pair) to align their action selection with the emotional cues. Also, we examined whether the absence of a direct behavior-regulating effect of the negative cue in Study III was due to the negative cue itself or to the concurrently available positive cue masking the negative cue’s potential effect. To this end, we split the demonstration of the two action-emotion pairs across two trials. In each trial, one action was thus demonstrated and emoted (PE, NE action), and one action was not demonstrated and un-emoted (UE action). For trial 1, we predicted that infants, who observed a PE action demonstration, would selectively imitate the PE action, whereas infants, who observed a NE action demonstration would selectively emulate the UE action. As to trial 2, we expected the complementary action-emotion pair to provide additional clarifying information as the model’s emotional evaluation of both actions, which should either lead to adaptive perseveration (if infants’ action selection in trial 1 had already drawn on the emotional cue) or adaptive change (if infants’ action selection in trial 1 signaled a disregard of the emotional cue). As to trial 1, our findings revealed that, as expected, infants imitated the PE action more often than they emulated the UE action. Like in Study III, this selectivity arose from an increase in infants’ propensity to perform the PE action from baseline to trial 1. Also like in Study III, infants performed the NE action about equally often in baseline and trial 1, which speaks against a direct behavior-regulating effect of the negative cue also when presented in isolation. However, after a NE action demonstration, infants emulated the UE action more often in trial 1 than in baseline, suggesting an indirect behavior-regulating effect of the negative cue. Yet, this indirect effect did not yield a selective emulation of the UE action, because infants performed both action alternatives about equally often in trial 1. Unexpectedly, infants’ action selection in trial 2 was unaffected by the emotional cue. Instead, infants perseverated their action selection of trial 1 in trial 2, irrespective of whether it was adaptive or non-adaptive with respect to the model’s emotional evaluation of the action. It seems that infants changed their strategy across trials, from an initial adherence to the emotional (particularly positive) cue, towards bringing about a salient action effect (Marcovich & Zelazo, 2009). In sum, Studies III and IV indicate a dynamic interplay of different action-selection strategies, depending on valence and presentation order. Apparently, at least in infancy, action reconstruction as one basis for selective action performance reaches its limits when infants can only draw on indirect intention-related information (i.e., which action should be avoided). Overall, our findings favor the action-reconstruction account (Csibra, 2007), according to which actions are flexibly organized along a hierarchy, depending on inferential processes based on extra-motor intention-related information. At the same time, the findings question the direct-matching hypothesis (Rizzolatti & Craighero, 2004), according to which the identification (and pursuit) of action goals hinges on a direct simulation of another’s behavioral cues. Based on the studies’ findings, a preliminary working model is introduced, which seeks to integrate the two theoretical accounts by conceptualizing the routes that activation induced by social cues may take to eventually influence an infant’s action selection. Our findings indicate that it is useful to strive a differentiated conceptualization of communicative cues, because they seem to operate at different places within the process of cue integration, depending on their potential to convey direct intention-related information. Moreover, we suggest that there is bidirectional exchange within each compound of adjacent sub-processes (i.e., between attention and encoding, and encoding and response), and between the compounds. Hence, our findings highlight the benefits of a multi-measures approach when studying the development of infants’ social-cognitive abilities, because it provides a more comprehensive picture how the concerted use of social cues from different domains influences infants’ processing and selection of instrumental actions. Finally, this thesis points to potential future directions to substantiate our current interpretation of the findings.. Moreover, an extension to additional kinds of coherence is suggested to get closer to infants’ everyday-world of experience.