Zeugenaussagen vor Gericht – über die Wahrheit und die Wahrheit über den Irrtum
Dr. Frank Maurer
Kern jeder richterlichen Tätigkeit ist, Zeugen zu hören und deren Aussage zu bewerten, Wahrheit zu erkennen, Lüge aufzudecken und Irrtum auszuschließen. In fast jedem Prozess werden Zeugen gehört, Zeugen sind das am meisten verwendete Beweismittel. Das Problem aber ist: Menschen sind nicht dafür gemacht, gute Zeugen zu sein. Ihr Auge ist keine Videokamera, ihr Gehirn kein Videorekorder. Der Mensch ist - soweit er als Zeuge auftritt - eine biologische Fehlkonstruktion.
Dieser MindTalk wirft einen Blick hinter die Kulissen bei Gericht und zeigt praxisnah, was Zeugen überhaupt leisten können und was Wahrnehmung, Aufmerksamkeit, Selektion und beschränkte Simultankapazität damit zu tun haben.
Attention recognition: Key to adaptive cognitive systems
Prof. Dr. Tanja Schultz und Dr. Felix Putze
The digital revolution is changing our world. Among the various promises of the near future are articifial intelligent technologies that provide just the right assistance when we need it. However, current digital assistants like navigation systems make us realize how difficult it is to strike a balance between support and distraction.
Getting this right requires technical cognitive systems that observe and interpret our daily activities, ultimately providing support when needed, while keeping our focus on the task despite natural and technological distractions.
In our talk, we will describe research and development at the Cognitive Systems Lab (CSL) towards such adaptive cognitive systems. We will show how modeling attention from neural and physiological data can help to create adaptive systems and will discuss several facets of attention which play a crucial role in everyday life. Furthermore, we will argue that humans usually rely on ambiguous cues and therefore cognitive systems should as well, processing a range of biosignals and integrate multiple modalities over time to reliably estimate a user's attention.
Several end-to-end systems and applications will be described that were developed within the framework of collaborative projects such as SmartHelm and EASE, in which the CSL team explores biosignals from speech, muscle and brain activities using machine learning methods to interpret user states and traits in everyday situations.
Neuronal architectures for affective brains
Dr. Wulf Haubensak
Neuroscience is undergoing two revolutions: circuit technologies allow to experimentally deconstruct neuronal network mechanisms of behavior, at the same time large brain and genomic databases create opportunities for computational mining of the neurogenetic organization of the brain. We fuse both approaches to explore the emotional brain, from genetic to systems levels.
Brains generate internal models of the world to interpret and guide interactions with the environment. Using circuit neuroscience, we delineated a cortico-limbic network that encodes subjective stimulus salience (‘how important?’) and affective valence (‘good or bad?’) at different hierarchies. We find that bottom-up salience signals in the amygdala instruct bodily feedback from the insular cortex to control affective memory and behavior. This network recruits ‘gut feelings’ into decisions, particularly when knowledge is scarce and integrates spatial signals to safeguard environmental interactions.
From the past 60 million years to the emergence of human societies, human ancestors faced increasing complex habitats, requiring ever more elaborate behavioral strategies. Computational reconstruction of human cognitive evolution allows to trace neurogenetic signatures of functional selection in affective networks and puts the circuit mechanisms identified above into evolutionary context.
Similarity-based processes in judgement and decision making
Prof. Dr. Bettina von Helversen
Similarity plays a fundamental role in how people make sense of the world. People group similar objects together and use similarity to infer how to respond to unknown objects or situations. Specifically, research on category learning and generalization assumes that people categorize new objects based on their similarity to past instances activated in memory. In addition to similarity-based processes, however, people also establish rules. These rules denote the relationship between an objects’ attributes and group membership (i.e. animals that fly tend to be birds) in categorization decisions or a quantitative criterion (i.e. larger animals tend to have higher lifespans) in quantitative judgment. Similarity and rule-based processes are often treated as two independent strategies or modules people recruit depending on the affordances of the task. However, evidence suggests that both processes interact. In the talk I give an overview of our research on how similarity and rule-based processes jointly determine responses in judgments and categorization decisions and how (dis)-similarity may fuel the formation of rules.
Cortical mechanisms for visual perception and restoring them in blindness
Prof. Dr. Pieter Roelfsema
I will argue that early visual cortex plays a crucial role in visual cognition – i.e. in tasks where we reason about what we see. Early visual cortex acts as a cognitive blackboard for read and write operations by higher visual areas, which can thereby efficiently exchange information. Inhibiting these interactions gives rise to selective deficits in visual perception. Elementary processes such as contrast detection are unimpaired. However, more complex tasks, which depend on the segregation of a figure from the background are impaired. Our results inspire new approaches to create a visual prosthesis for the blind, by creating a direct interface with the visual cortex. I will discuss how high-channel-number interfaces with the visual cortex might be used to restore a rudimentary form of vision in blind individuals.
Where are the switches in the brain? Neural mechanisms of selective attention in experiment and theory
Prof. Dr. Andreas Kreiter und Dr. Udo Ernst
Human and non-human primates' brains are capable of performing a wide variety of quickly changing, often attention-dependent tasks in complex environments. Unlike digital computers, they do not have a software, i.e., different programs executed by the same central processing unit, which are called and executed according to current needs. Instead, complex networks of neurons throughout the brain can serve different purposes of information processing. To perform different tasks at different times, these networks need to be reconfigured accordingly at a time scale of seconds or less. However, networks of neurons do not have switch-like elements to change their connectivity pattern. Using a combination of experimental and theoretical results, we will argue that states of oscillatory synchronization between changing subsets of neurons modulate the pattern of effective connectivity within neuronal networks. We will show that such dynamic changes result in different functional circuits that route information depending on the behavioral task and selective attention, and discuss (putative) control mechanisms which are capable to establish and maintain appropriate network configurations.