TU Berlin

Cognitive Modeling in dynamic Human-Machine SystemsZiying Zhang


Page Content

to Navigation

Ziying Zhang


Master's Student



Marchstr. 23

10587 Berlin

Room: MAR 3.039


Consultation hours: on appointment



Deciphering dynamic decision making using cognitive modeling including
subjective assessment

Decision-making is a high-level and interdependent cognitive process, which includes but not limited to perception, attention, and memory. People make a decision usually based on their previous experience and the feedback from the changing environment. Thus, the decision-making process is always dynamic, which means more complex to predict and occur in real-time.

The Instance-Based Learning Theory (IBLT) is a theory, developed by Cleotilde Gonzalez,etc., to explain how humans make decisions in dynamic tasks. People make decisions relying on their accumulated experience and retrieving past solutions for similar cases stored in memory. Thus, the accuracy of decision can only improve through more times interaction with similar situations gradually. But for many learning tasks in real life, they usually already have some set rules needed the participants to follow. Thus, the combination of both instance-based learning and rule-based learning is the foundation of decision making for those learning tasks.

In this research, we choose a complex rule-based category learning task- to classify or distinguish the sounds- as the research target. Participants need to identify the conjunction of two rules which defined a target category firstly, then they learn repeatedly and finally adapt to a reversal of feedback contingencies.

We applied the method of cognitive modeling to understand this dynamic decision-making process better. An ACT-R model, based on the Lisp language, is developed for the core aspects of this task as well as the main steps of simulating the thinking process. It solves the categorization task by first trying out one-feature strategies and then switching to two-feature strategies as a result of repeated negative feedbacks.

This thesis work is based on previous work which Sabine had published. And it mainly does two aspects improvement: one is to develop a more accurate model mainly considering participants' pre-knowledge of the correct button (which they use to express their final decision) when learning from feedback. The other one is to add the control group with questionnaire/survey in the middle of the new experiment, to observe and compare the influence of the new subjective assessment.


Research interests

Product and interaction design, User assessment, Cognitive process



Human Computer Interaction and Design (HCID) (M.Sc)

  • EIT Digital Master School (2015 – present)
    • Technische Universität Berlin (TU Berlin) (2016 - present)
    • KTH Royal Institute of Technology (2015-2016)

Macromolecular Materials and Engineering (B.Sc)

  • Fudan University, P. R. China (2008 – 2012)
  • Thesis: Study on the best condition of Gelatin-Chitosan composite material’s preparation

Chemical Engineering (Exchange student)

  • National Tsing Hua University (NTHU), Taiwan (2010)


Work Experience

Product Manager (Full-time), Tencent Inc. (SEHK700) (2012-2014)

  • An intermediate level product manager in charge of both mobile and web products
    • Tencent iCare for iPhone (V1.6 -V2.1)
    • Tencent Weibo (2013 Version, Weibo Badge and Daily Task)
  • The work responsibility includes product design, interaction design, user and content operation, and project management.

Thesis Student, Key Laboratory of Molecular Engineering of Polymer, Fudan University (2012)

  • Summer Internship, Tencent Inc.(2011)
  • Student Assistant of psychological consultant, Mental Health Center, Fudan University (2011)




Quick Access

Schnellnavigation zur Seite über Nummerneingabe