Computational Affective Modelling I

Fr.o.m.

31.7.2025

2 sp

Fördjupade studier

This course presents a introduction to modelling of affective phenomena. It has a particular focus on theories of human suffering, or mental pain. The basic premise is that we can use the modern theory of artificial intelligence (AI) to model and understand human suffering or mental pain. Both humans and sophisticated AI agents process information about the world in order to achieve goals and obtain rewards, which is why AI can be used as a model of the human brain and mind, and this includes affective phenomena. This course intends to make the theory accessible to a relatively general audience, requiring only some relevant scientific background (see prerequisites); it is based on the e-book available at https://www.cs.helsinki.fi/u/ahyvarin/painintl/(Öppnas i en ny flik) .

Part I begins with the fundamental assumption that human suffering is mainly caused by frustration. Frustration means the failure of an agent (whether AI or human) to achieve a goal or a reward it wanted or expected. Frustration is inevitable because of the overwhelming complexity of the world, limited computational resources, and scarcity of good data. The concept of threat, related to risk and fear, provides a complementary approach to modelling suffering. Fundamental in such modelling is the idea of learning, or adaptation to the environment. While AI uses machine learning, humans and animals adapt by a combination of evolutionary mechanisms and ordinary learning. Another important concept here is the dual-process nature of human cognition: the division into logical-symbolic and neural-network information processing.

Part II explores in more detail various aspects and limitations of intelligent agents and learning algorithms, and the implications of such limitatios regarding suffering. In particular, such limitations imply that an agent acting in the real world must cope with uncontrollability, unpredictability, and uncertainty, which all lead to frustration. In the end, the computational theory is used to derive various interventions that should reduce suffering in humans. The amount of frustration is expressed by a simple equation which indicates how it can be reduced. The ensuing interventions are very similar to those proposed by Buddhist and Stoic philosophy, and include mindfulness meditation.

Tilläggsuppgifter och anmälan

Pris och anmälningstid

Kontrollera informationen på högskolans webbsida

Gå till högskolans webbplats(Öppnas i en ny flik)

Detaljerad information:

Tidpunkt

Fr.o.m.

31.7.2025

Utbildningsområden

Databehandling och kommunikation

Omfattning

2 sp

Kod

DATA20060

Förkunskapskrav

Lärandemål

After Part I of the course, the student

- has become acquainted the basic concept of modelling affective phenomena, such as suffering, using the theory of autonomous agents and machine learning

- understands the importance of frustration in modelling human suffering, and how it can be modelled in an AI agent

- understands how threat and risk can be modelled in a probabilistic AI framework

- understands the idea of dual processing in a cognitive system and its implications for action selection and affective phenomena

After Part II of the course, the student additionally

- understands how limitations in information-processing capacity and the amount of data used in learning are related to affective phenomena

- In particular, is able to discuss the implications of uncertainty and uncontrollability in agent design, and their relation to affective phenomena and emotions

- has a general understanding of how interventions can be designed to reduce negative affect

- has the expertise necessary to use these ideas and theories in areas such as HCI, computational neuroscience, or safe AI.

Studiematerial

Study materials: All material is provided in the MOOC, which includes text pages, multimedia content (e.g. short recorded lectures), and multiple-choice questions (the last serving both for learning and assessment).

Literature: While the MOOC contains everything, it is quite precisely based on an e-book written by Aapo Hyvärinen, see https://www.cs.helsinki.fi/u/ahyvarin/painintl/(Öppnas i en ny flik) . The factual contents of the MOOC and the e-book are basically equivalent, except that some details are left out of the MOOC, while multimedia content will be created for the mooc merely as pedagogical devices without increasing the factual content.

Tilläggsuppgifter

Completion methods: Independent study (MOOC)

Assessment practices and criteria: Assessment based on multiple-choice questions in the MOOC system.

Activities and methods in support of learning: This is a MOOC and based on independent study of material on the MOOC web pages. These contain text pages, multimedia content (e.g. short lectures), and multiple-choice questions.

Target groups: Primarily, advanced students (starting from 3rd year Bachelor’s) of data science and computer science. Secondarily, students in any related disciplines since the prerequisites are kept quite low, especially International Bachelor’s of Science and Liberal Arts and Sciences (both in 3rd year). The course is not obligatory in any program. Being a MOOC, it is open to “everybody” under the usual conditions of UH MOOCs.

Teaching period: This MOOC will be available any time. The goal is to make it available at the beginning of academic year 2025-26.

Recommended time or stage of studies: 3rd year Bachelor's of CS, or any year of DS/CS master. For non-CS students, typically an optional course in a Master’s degree, but in International Bachelor’s in Science and Liberal Arts and Sciences, advanced (e.g. 3rd year) Bachelor’s students.

Study module: This is part of the specialization studies module "Cognition in Brain and Machines" in the data science master’s program.

Expiry of studies: to be decided

Language of Instruction: English

Arrangör

Helsingin yliopisto

Tilläggsuppgifter

avoin-student@helsinki.fi