Themen für Masterarbeiten

Beschreibung: Zahlreiche Studien haben bisher versucht, kognitive Inhibition als Fähigkeit zu messen. Faktoranalytische Methoden (z.B. Strukturgleichungsmodelle) wurden auf Leistungsmasse diverser experimenteller Inhibitions-Paradigmen, wie z.B. Stroop-Aufgabe oder Eriksen Flanker-Aufgabe, angewandt. Diese Studien haben inkonsistente Ergebnisse hervorgebracht: In manchen Studien korrelieren die Inhibitionsmasse positiv miteinander, so dass ein gemeinsamer Faktor "Inhibition" identifiziert werden konnte, aber in vielen Studien gelang das nicht. Ein Grund dafür könnte sein, dass die experimentellen Effekte, durch die die Fähigkeit zur Inhibition geschlossen wird (z.B. Flanker-Effekt), recht klein sind, so dass sie schwer reliabel zu messen sind. Ziel der Masterarbeit ist, eine Reihe neuer Inhibitionsaufgaben (oder Varianten existierender Aufgaben) zu prüfen, bei denen erwartet werden kann, dass die Inhibitionseffekte grösser sind, so dass sie als bessere Masse der kognitiven Inhibition in Frage kommen.
Kontakt: Prof. Dr. Klaus Oberauer, E-Mail

Beschreibung: Wie viele Ereignisse in unserer Umwelt können wir gleichzeitig beachten? Gibt es dafür eine begrenzte Kapazität? Ziel der Masterarbeit ist, ein Mass für die Kapazität der visuellen Aufmerksamkeit zu entwickeln und zu überprüfen.
Kontakt: Prof. Dr. Klaus Oberauer, E-Mail

Beschreibung: Cognitive measurement models such as signal detection theory, the memory measurement model M3, or the diffusion model estimate psychologically meaningful parameters, like memory strength, response caution, or the rate of evidence accumulation, from behavioral data. These models are increasingly fit in a hierarchical Bayesian framework that estimates parameters for each person while borrowing strength across the sample, for example with the bmm R package. Before we interpret such parameters, however, we need to know that they can actually be recovered: that the values we estimate match the values that generated the data, and how many participants, trials, and conditions this requires. At present, recovery studies are run ad hoc and look different for every model, and there is no agreed standard for how to conduct and report them.

The proposed master thesis develops a principled, standardized workflow for parameter recovery and simulation-based calibration of the hierarchical measurement models in bmm, starting with fast-estimating models and extending to slower ones as computation allows. The student simulates data from known parameters across a grid of design factors, namely the number of participants, trials per person, within-subject conditions, and effect sizes, refits the models, and quantifies recovery through recovery correlations, bias, coverage, and simulation-based calibration. The result is a set of reusable tools and concrete data-requirement recommendations that tell researchers how much data they need before the parameters of a given model can be trusted. The work is self-contained and needs no data collection, which suits a methodologically minded student who enjoys simulation and programming in R.
Kontakt: Dr. Gidon Frischkorn-Bartsch, E-Mail

Beschreibung: Complex span tasks, often used to investigate working memory, combine storage demands, such as remembering memoranda in serial order, with processing demands like judging whether a math equation is correct or incorrect. Interference theories of working memory assume that processing imposes additional interference and thus decreases the fidelity of representations, leading to worse performance compared to memory tasks not involving processing. Yet, it is unclear with which aspect of memory representations in working memory the processing of distractors interferes specifically.

The proposed master thesis should investigate this using adapted complex span task, in which participants remember feature pairs (such as digit-color pairs, or color-spatial location pairs), and the distractor task involves processing the same kind of pairs. Using this task, we want to investigate in how far processing one of the features (e.g. color or spatial position) affects the interaction and fidelity of memory representations during retrieval depending on using the processed or non-processed feature as retrieval cue. This might be a promising way to evaluate in how far processing specifically interferes with feature representations or bindings in WM.
Kontakt: Dr. Gidon Frischkorn-Bartsch, E-Mail

Beschreibung: Selective attention tasks such as the Stroop task and the Simon task produce effects that are among the most robust in cognitive psychology from a group-level perspective. From an individual-differences perspective, however, those effects fare much more poorly. Part of the reason is that measurement noise remains high with the number of trials typically used in those tasks, leading to poor reliability and attenuated cross-task correlations (the so-called reliability paradox). Another, less appreciated, part of the reason is that little theoretical analysis typically goes into designing selective attention tasks. The result is that the effects being compared across tasks often reflect multiple processes for each of which individuals might vary to some extent independently. The present thesis will attempt to solve these problems by collecting a number of trials large enough to minimize measurement noise and by designing some of the most popular selective attention tasks so as to dissociate the processes potentially contributing to individual variation.
Kontakt: Dr. Giacomo Spinelli, E-Mail

Beschreibung: Selective attention typically allows us to process and respond to a relevant stimulus (or stimulus component, such as the ink color of a colored color name) even in the presence of a more salient stimulus (or stimulus component, such as the color name itself). However, such situations typically produce a performance cost (such as the Stroop effect), suggesting that selective attention is capacity-limited. In the cognitive psychology literature, selective attention's capacity limits have typically been explained by assuming that there is a limited amount of cognitive resources that we can use. However, while such a resource-based model predicts increased interference effects in resource-demanding situations (e.g., when a difficult secondary task is required), this result pattern has not always emerged in recent years. In parallel, there has been a growing interest in the role of motivation in selective attention and cognitive control more broadly, with some models assuming that effects of cognitive effort might more typically emerge in motivational tradeoffs than in performance costs. Building on these ideas, the present thesis will involve experimental work to re-examine the capacity limits of selective attention and determine whether they are subject to motivational tradeoffs.
Kontakt: Dr. Giacomo Spinelli, E-Mail

Beschreibung: Mind-wandering, the drift of attention away from the task at hand, is by its nature a momentary state, and it is typically captured in the moment with thought probes that interrupt a cognitive task and ask people where their attention just was. What a questionnaire like the BMW-3 measures is something different: a person's general disposition to let their mind wander, reported in the abstract rather than in the moment. Such self-report dispositional measures are convenient and widely used, but their validity is still debated, precisely because it is unclear how well a summary judgement of one's own tendency to mind-wander corresponds to how often the mind actually wanders during a task. Establishing that link is what would give a scale like the BMW-3 its validity as a trait measure.
The proposed master thesis explores and extends this validation by asking whether performance-dependent thought probes can clarify the relationship between self-reported mind-wandering disposition and mind-wandering reported in the moment. As a starting point, the student can draw on the Ambizione dataset, in which the BMW-3 was administered on two occasions, to examine the stability of the dispositional measure and its relations to working memory capacity, fluid intelligence, and reaction-time variability as a behavioural marker of lapsing. The core of the thesis is a newly collected online study that embeds thought probes in a cognitive task and compares two probing schedules: probes delivered at fixed intervals, and adaptive probes triggered by live fluctuations in performance, such as spikes in reaction-time variability that behaviourally signal a lapse. The question is whether performance-dependent probing yields in-task mind-wandering reports that align more closely with the BMW-3, and thus offers a sharper test of whether the questionnaire measures what it claims to. The study is deployed with jsPsych via JATOS and Prolific, and suits a student who wants to combine data collection with a methodologically novel measurement idea.

Kontakt: Dr. Gidon Frischkorn-Bartsch, E-Mail

Beschreibung: Multitasking is typically theoretically constrained by a central bottleneck, resulting in dual-task costs. Yet, it remains debated whether this reflects an immutable structural limitation or an adaptive optimization of processing strategies. From an adaptive perspective, a serial strategy should be preferred when minimising cross-task confusion is critical (Logan & Gordon, 2001). Conversely, a less effortful parallel processing strategy can also be adopted to increase the efficiency, reflecting a trade-off between optimizing the performance and minimizing the effortful processing effect (Lehle et al., 2009). Importantly, the dual cost can be mitigated by prolonged stimulus onset asynchrony (SOA, Janczyk, 2013), which reveals parallel processing in pre-central stages, and by reducing cross-talks between stimuli and responses (Hommel, 1998), or even practise (Oberauer & Kliegl, 2004), enabling the transition to parallel processing. The goal of this master thesis is to investigate factors that determine strategic preference, focusing specifically on the SOA duration and frequency distributions, as well as explicit instruction.
Kontakt: Dr. Shuangke Jiang, E-Mail

Beschreibung: Multitasking is ubiquitous in everyday life, despite performance declines under multitasking compared to single-task settings. For example, when alternating between two different tasks in rapid succession, responses are slower and less accurate than when ony one task is performed over and over again. Originally, this was interpreted as a wroking memory (WM) load effect: when alternating between two tasks more tasks have to be maintained in WM. However, perhaps surprisingly, alternating between three tasks does not worsen performance compared to alternating between two tasks, which contradicts the WM-load explanation. It was suggested that tasks are not maintained in WM, rather only the current task is, and the others are offloaded to long-term memory, which is not capacity limited. In this empirical thesis, we will conduct an experiment to put these opposing explanations to a test. We will compare performance when participants alternate between two or three tasks. By changing the tasks often, we limit the opportunity for participants to learn the tasks, hence making it more difficult offload tasks to long term memory. This will help understanding to what extent WM capacity limits contribute to the worse performance in multitasking compared to single tasking.
Kontakt: Dr. Elena Benini, E-Mail

Beschreibung: Das Arbeitsgedächtnis hält Information kurzfristig für die Verarbeitung bereit. Wir nehmen an, dass all unser Denken darin besteht, dass wir mentale Repräsentationen im Arbeitsgedächtnis manipulieren. wenn wir eine Reihe von Wörtern im Arbeitsgedächtnis halten, nicht nur deren gedrucktes Aussehen oder deren Sprachklang repräsentieren, sondern auch die Bedeutung der Wörter. Überraschenderweise sprechen einige neuere Befunde dagegen, dass die Bedeutung von Wörtern im Arbeitsgedächtnis repräsentiert wird: Wenn Personen sich mehrere Wörter kurzfristig merken müssen, die sich in ihrer Bedeutung ähnlich sind, verwechseln sie diese Wörter nicht häufiger miteinander, als wenn sich die Wörter unähnlich sind. Ziel der Masterarbeit ist, zu untersuchen, ob die Bedeutungsähnlichkeit eine Rolle spielt, wenn wir uns Bilder alltäglicher Objekte merken müssen. Daraus können wir schliessen, ob die Bedeutung von Objekten im Arbeitsgedächtnis repräsentiert wird.
Kontakt: Prof. Dr. Klaus Oberauer, E-Mail

Beschreibung: Ein Befund aus der Forschung zum Arbeitsgedächtnis ist "output interference": Wenn mehrere Elemente einer Liste, die im Arbeitsgedächtnis gehalten wird, nacheinander getestet werden, dann beeinträchtigt jeder Test die Erinnerung an die noch verbleibenden Elemente, so dass von einem Test zum nächsten die Erinnerungsleistung schlechter wird. Ziel der Masterarbeit ist, herauszufinden, was diesen Effekt verursacht: Ist es das Abrufen eines Elements aus dem Arbeitsgedächtnis, oder entsteht der Effekt durch die Antwort, die man beim Test geben muss (z.B. das getestete Element reproduzieren)?
Kontakt: Prof. Dr. Klaus Oberauer, E-Mail

Beschreibung: New information attempting to enter working memory (WM) must go through a period of consolidation, which stabilizes the representation for storage. This consolidation process appears to be dependent on temporal and attentional constraints. While well studied, several lingering questions remain about the structure of information during this consolidation process and the nature of the mechanism. This project aims to use a relatively new form of visual masking, object-substitution masking, to probe the consolidation process. Object-substitution masking acts by using a sparse visual mask (usually a few surrounding dots) that linger after the removal of a target object. This sparse trailing representation appears to ?remove? or ?substitute? the target object representation in memory without affecting the perceptual contents of the original image. Using this procedure, we will (1) attempt to determine a fine timescale upon which the representation is substitutable during consolidation and (2) examine the nature of response errors for items that have been replaced. The goal of this project is to gain a greater understanding of visual working memory formation, separate from the underlying perceptual precursors.
Kontakt: Dr. Jacob Zepp, E-Mail

Beschreibung: Eine Aufgabe der neuropsychologischen Testung ist, zu diagnostizieren, ob eine Person eine tatsächliche kognitive Einschränkung hat oder eine solche nur simuliert (um etwa die Versicherung zu betrügen). Ein Ansatz dazu ist, die Schwierigkeit von Items in einem kognitiven Test so zu variieren, dass es wenig offensichtlich ist. Personen mit echter Beeinträchtigung sollten einen Schwierikgietseffekt zeigen; Simulanten sollten diesen Effekt nicht zeigen. Ziel der Masterarbeit ist, einen solchen Test zu konstruieren und anhand einer Stichprobe von gesunden Personen zu testen.

Kontakt: Prof. Dr. Klaus Oberauer, E-Mail

Beschreibung: The consolidation, or stabilization, of visual information into working memory (WM) is gated by an attentional selection process. When items are presented sequentially, one at a time, it is relatively simple to determine the order of item entrance into WM. When multiple items are presented simultaneously, all at once, this order becomes much less clear. This lack of clarity in the simultaneous case has led to debate on the nature of the consolidation process. Can multiple items be processed in parallel, or do they still enter WM serially even when all available at once? This project will aim to experimentally address this debate through the controlled direction of attention in simultaneous visual presentations, and the evaluation of the impact of this manipulation on successful storage.
Kontakt: Dr. Jacob Zepp, E-Mail

Beschreibung: Working memory (WM) is crucial for cognitive functions such as problem-solving, learning, and decision-making. Training in WM tasks can enhance cognitive performance, with improvements sometimes transferring to similar tasks?a phenomenon known as near-transfer. However, the mechanisms driving this transfer remain unclear. Do gains arise from shared procedural knowledge, common strategies, or both? This study aims to exam the extent to which commonly used WM training tasks rely on overlapping procedural knowledge or strategic approaches and how these shared components contribute to near-transfer effects. By comparing performance across similar WM paradigms and stimulus types, we will assess the degree of structural and strategic overlap between tasks. Identifying these core components will deepen our understanding of the cognitive architecture of WM and the mechanisms underlying transfer.
Kontakt: Dr. Shuangke Jiang, E-Mail

Beschreibung: Das Arbeitsgedächtnis wird oft als eine begrenzte Ressource beschrieben, die sowohl für das kurzfristibe Behalten von Information als auch für das Verarbeiten erforderlich ist. Wenn während einer Behaltensaufgabe eine Verarbeitungsaufgabe erledigt werden muss, sollte die Geschwindigkeit (und evtl. die Genauigkeit) der Verarbeitungsaufgabe davon abhängen, wie viel Information gleichzeitig im Arbeitsgedächtnis gehalten wird (also vom "memory load"). Einige Studien sprechen dafür, dass dies der Fall ist: Bei grösserem "memory load" werden die Reaktionszeiten einer einfachen Verarbeitungsaufgabe langsamer. Es gibt aber Hinweise darauf, dass dies nicht die Belastung einer Ressource widerspiegelt, sondern die Erwartung der Person, dass die Aufgabe insgesamt schwieriger sein wird. Ziel der Masterarbeit ist, zwischen diesen Alternativen zu entscheiden. Diese Masterarbeit eröffnet die Möglichkeit, die Daten mit Hilfe eines einfachen mathematischen Modells (einer vereinfachten Variante des Diffusionsmodells) auszuwerten. Da diese Arbeit neben experimenteller Forschung auch die statistische Modellierung der Daten beinhaltet, ist eine gewisse Affinität für statistische Methoden Voraussetzung.
Kontakt: Prof. Dr. Klaus Oberauer, E-Mail

Beschreibung: There is a limit to the information that we can process and maintain simultaneously. To date, it remains unclear what causes this limit. For example, can we maintain more information if the different items to memorize overlap by some dimensions (e.g., colour, shape)? This would reduce the total amount of information to memorize. Also, it may help creating groups (i.e., chunks) of items. Or does overlap increase interference between the different items, which become more confusable? In this empirical thesis, we will conduct an experiment to test whether WM performance increases or decreases when items to memorize overlap by some dimensions, compared to when they are distinct. This will help understanding whether we can use overlapping information to reduce WM load or whether the resulting similarity between items imposes higher demands on WM.
Kontakt: Dr. Elena Benini, E-Mail