Time: Every other Friday from 20 January until 28 April 2023, 14:15–15:45 EET
Place: University of Helsinki main building, Unioninkatu 34, 4th floor, room U4078, and online (Zoom link)
Learn more about the presenters and their topics below.
In everyday life we carry out many tasks that are apparently easy and simple, but are in fact underpinned by surprisingly sophisticated cognitive and neural mechanisms. This impressive adaptability of human perception and cognition is starkly revealed, for example, in artificial intelligence and robotics where everyday tasks have turned out to be huge challenges.
To understand why we can have a tremendous human advantage for these sorts skills, yet “superhuman AI” for others, we need to understand how the human mind and brain cope with the complexity and ambiguity of real-world tasks. This can only be achieved by taking a properly ecological approach to how we do research in cognitive science, both at the level of theory and methodology. But what does an “ecological” approach amount to, and what counts (and does not count) as an “ecological” approach to Cognitive Science?
In this talk, I present ten simple rules to evaluate the “ecological” credentials of empirical research in the cognitive sciences, discuss their rationale, and the ways we researchers all too often violate them.
Otto Lappi is an Academy Research Fellow and a Senior Lecturer in Cognitive Science at the University of Helsinki. He is interested in all aspects of the cognitive basis of everyday & expert performance – driving performance most especially. He leads a lab at the University of Helsinki, and his empirical work is mostly based on eye tracking & experimental work in challenging naturalistic tasks, especially combining observations "in the wild", simulator studies, and controlled lab tasks.
Besides his empirical research, he also has a background in philosophy, where he is interested in the conceptual and methodological foundations of cognitive science.
This talk is based on the recent book with the same title, freely available at www.painfulintelligence.info . The book uses the modern theory of artificial intelligence (AI) to 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. This book intends to make the theory accessible to a relatively general audience, requiring only some relevant scientific background.
Aapo Hyvärinen studied undergraduate mathematics at the universities of Helsinki (Finland), Vienna (Austria), and Paris (France), and obtained a Ph.D. degree in Information Science at the Helsinki University of Technology in 1997. After post-doctoral work at the Helsinki University of Technology, he moved to the University of Helsinki in 2003, where he was appointed Professor in 2008, at the Department of Computer Science. From 2016 to 2019, he was Professor of Machine Learning at the Gatsby Computational Neuroscience Unit, University College London, UK. Aapo Hyvärinen is the main author of the books "Independent Component Analysis" (2001), "Natural Image Statistics" (2009), and "Painful Intelligence" (2022).
This talk sends up a trial balloon of some semi-raw thoughts about the human mind. Assuming that reality exists independently of our knowing and perceiving, I will first claim that we all live in individual “caricature worlds” that are shaped by our brains and senses, prior experiences, and sensorimotor and emotional bodily states so that we automatically emphasize salient and surprising events. We can share the caricature worlds to some extent because we live in similar environments and are interacting with other people in the society.
Second, I will argue that the attempts to solve the “mind–body (or mind–brain) problem” would greatly benefit from more serious consideration of the evolution and emergence of brain functions, the special features of living matter, and social interaction.
Throughout the talk I will emphasize the importance (even primacy) of motor action, and thereby of the body, for the development and proper functioning of the human mind, including thinking.
And finally: Understanding the human mind is a wicked problem that does not fall into the arms of any single discipline. We thus should strive for convergence research where researchers with different backgrounds share their target problems.
Riitta Hari, MD PhD, is Academician of Science and Distinguished Prof. (emerita) of Human Systems Neuroscience and Brain Imaging at Aalto University, Finland. Hari has studied sensory, motor and cognitive functions in healthy humans and patient groups, with main focus in the dynamics of brain function. She has advocated “two-person neuroscience” for the study of the brain basis of social interaction, and she is currently trying to bridge art and neuroscience. Hari was the initiator and chair of Mind Forum (funded by the Finnish Cultural Foundation) and the lead author of ‘Ihmisen mieli’ (‘The Human Mind’; Gaudeamus 2015).
It has been claimed—perhaps most famously by Bertrand Russell—that the concept of cause has no place in the empirical sciences. This is arguably an exaggeration, but with a grain of truth. Undeniably, it has not been easy to see how philosophical accounts of causal connections fit with the scientific understanding of particular physical phenomena. This is understandable for so-called standard approaches to causation (i.e. in terms of regularity, conditions, counterfactuals, and intervention), since they are arguably attempts to elucidate how people think and reason about causation rather than attempts to figure out what causation really is. However, in the last 30 years or so there has been a surge of interest in developing realist accounts of causation that are informed by—or at least compatible with—the theories and findings of the empirical sciences. In this talk I will present four realist accounts of causation and discuss whether they really succeed in explaining two kinds of physical phenomena—collisions between billiard balls, and how water dissolves salt—in a manner compatible with the received scientific understanding of said phenomena. The four accounts are: (i) transmission accounts, (ii) mechanistic accounts, (iii) powers-based accounts, and (iv) the powerful particulars account. I will argue that the first three accounts fare badly in this exercise, mainly because they all assume that physical influence is unidirectional, while the natural sciences insist all interactions are reciprocal. The notion of “reciprocity” is admittedly ambiguous, and I will outline three different senses in which it currently understood. I will argue that the fourth account fares much better than the other. This is not surprising, because it was developed in the first place to address the flaws of the other three and develop an understanding of causation in terms of reciprocal action.
Valdi Ingthorsson is a lecturer in philosophy at the University of Helsinki. His research centers on various issues in metaphysics, such as the nature of time, persistence, causality, powers, substances, processes, and truth. He is the author of the books McTaggart’s Paradox (2016) and A Powerful Particulars View of Causation (2021). He also has an interest in various issues in the philosophy of science, such as the difference between the natural and human sciences, and between quantitative and qualitative research methodology.
What is the future of AI in Europe and in the world? In March 2023, the European Parliament is expected to vote on the new regulation for artificial intelligence, the AI Act. The AI Act introduces harmonized rules for the design and deployment of AI systems across different sectors in Europe. During the lengthy political process the AI Act has become almost mythical and, unlike most technology regulation, it is broadly discussed among politicians, engineers, and civil society actors. This M&M talk addresses the myths and problematisations of AI regulation from the combined perspective of cognitive science and socio-legal studies on law, technology, and society. The discussion is divided into four sections. First, we address the so-called definition dilemma, which requires translation of technical understanding of AI into a precise application scope of new regulation. Second, we question the feasibility of different proposed solutions that range from transparency to human oversight of AI systems. Third, we examine the assumptions behind the EU’s risk-based approach towards technology regulation in general. And finally, we question the purpose and objectives of the AIA: do we really need AI-specific regulation in the first place? What would be the alternative ways to envision our AI-infused futures?
Anna-Mari Rusanen is a philosopher of artificial intelligence and cognitive sciences. Her research topics vary from the philosophical foundations of artificial intelligence to the societal implications of algorithmization, and from the nature of computational explanations to the representational accounts of cognitive systems. Her recent work focuses on the roles that paradigmatic examples play in current debates on the societal implications of AI, such as in the context of the AI Act. Currently Rusanen works as an university lecturer in cognitive science, (Department of Digital Humanities, University of Helsinki), and as a senior advisor on scientific, societal and ethical aspects of AI in Ministry of Finance (Finnish Governance).
In this talk, we will discuss how quantum software is different from classical software, the new challenges it poses, and the new opportunities it brings to different application domains: in our case in particular, to machine learning. The computational thinking required for quantum programs is very different from classical programming. Understanding, creating, and testing quantum software has a steep learning curve. New ideas, competencies, and abstractions are needed to make quantum computing accessible to experts in the ICT field. While research on quantum hardware is very active, the work on quantum software and algorithms is still in its infancy.
Jukka K. Nurminen is a professor of computer science at the University of Helsinki. He has worked extensively on software research in the telecom industry at Nokia Research Center, in academia at Aalto University, and in applied research at VTT. His key research contributions are on energy-efficient software, mobile peer-to-peer networking, and cloud solutions but his experience ranges widely from applied optimization to AI, from network planning tools to mobile apps, and from software project management to tens of patented inventions. He has e.g. led the Green Big Data project with CERN and many research activities on mobile phone and cloud energy consumption. Currently, his main interests are in the engineering of machine learning systems, combining data science and high performance computing, and software development for quantum computers.
Color perception is computationally hard because the light signal arriving the retina confounds information about surface color and illumination color. Without constraints, it is not possible to estimate the color of objects from this signal. Statistical regularities in natural surfaces and illuminants, learned through interacting with our environment, may contribute to our ability to solve this computational problem and estimate surface color. In this talk, I will discuss perceptual phenomena such as memory colors, central tendency bias, and color categorization to highlight the role of prior knowledge in color perception and color constancy.
Maria Olkkonen is a principal color scientist at Microsoft and an adjunct professor at the University of Helsinki. She received her PhD in color perception at the University of Giessen in 2009, after which she worked as a postdoctoral fellow at the University of Pennsylvania and at Rutgers University. She returned to the University of Helsinki in 2015 as an Academy Research Fellow and simultaneously worked as an assistant professor at Durham University in England. She joined Microsoft in 2021 to work on improving the image quality of AR displays. In her research, she has demonstrated the importance of prior knowledge and memory on color perception, investigated the neural underpinnings of object and color perception, and the development of color constancy in children. She is currently extending this research into AR/VR applications at Microsoft.