![Thumbnail](/bitstream/handle/20.500.11850/604304/230307_Thesis_Manuscript_Reviewed_PC.pdf.jpg?sequence=8&isAllowed=y)
Open access
Author
Date
2022Type
- Doctoral Thesis
ETH Bibliography
yes
Altmetrics
Abstract
It seems clear now that the main aspect of how one deal with the grasping of the whole world, including oneself, is of a probabilistic kind. And this is highly linked to the fact that the concept of information has gradually dominated our production of ideas since the 19th c. One must now deal with an overdue time of acceptance that embracing informational and communication models should be most pervasive and necessarily overcome to build a new ground of capacity, another boundary to explore. Such a statement is not to be confused with primitive positivist views but rather intends to express a belief in the models we elaborate to observe and understand ourselves as intelligent humans and, concurrently, how to model the world that surrounds us. The informational paradigm, its inferential and probabilistic approach to information, is taken from many fields involved in perception, cognition and representation of information at many levels and which I believe should provide more contemporary approaches to thinking of architectural modeling, the encoding of parts and their articulations.
Throughout the title: ‘Neurodesign, Modeling with Neural Potentials’, the term ‘neural potentials‘ means twofold: ‘potentials‘ as in electric potentials, vessels of information to be decoded, and ‘potentials‘ as in ‘capacity‘ held by such information. The thesis comprises three main chapters, its conclusion and an appendix containing publications and research materials produced along the research. Its general intent reflected throughout the chapters is to balance the research between empirical and theoretical findings to mutually inform each other on the prototyping of potentially novel ways to approach architectural modeling and its encoding.
The first chapter serves as an expanded ‘Introduction‘ into ground contexts and positions for this research. In order to emphasise peculiar circumstances which are laying a base for the main arguments of this thesis, the sections are organised along key notions which have been deprived of some of their constituents, in apophatic terms, and by looking mainly at the past two centuries. The chapter begins with a general introduction to the identified background contexts, the thesis structure, its methods and the theoretical positions which led to these directions. Followingly, the main section develops the ‘ground contexts‘ of ‘modeling intelligence on apophatic grounds‘ by emphasising on - the obsolescence of particular centralities of scientific thinking after a 20th c. crisis of intuition, and the opportunities revealed for the pursued thinking of computation in architecture (‘Knowledge without Center‘), - the transformation of reason into rationality and its reapparition in an objective formulation along the same period (‘Intelligence without Reason‘), - the capacity of models of computation in their incompleteness and complementarity as a potentially more capacious form of modeling with intelligence (‘Computation without Universality‘), - the computational models dealing with representational theories of the mind and the way some of its elements suggest ways to deal with human cognitive capacity to create non-exhaustive values from the world (‘Minds without Meaning‘), - the endless search for correlating spatially temporal and dynamic events of the mind which led to abundant ideas about the mechanisation of natural symbolic processing (‘Events without Places‘), - the overlooked aspect of temporal agency in architectural modeling in regards to the human mind and the way it relates to events and its valuation (‘Articulation without Representations‘).
These sections are then followed by a synthetic account of the corollary and authored precedents of this research for positing a retroactive reflection before summing the contextual introduction and leading to the main current research question on architectural modeling with neural potentials.
The second chapter (‘Modeling with Neural Potentials‘) brings the question forward by intending to develop a prototypical theory on architectural modeling with neural potentials. It does so by looking selectively into a broad spectrum of joined ideas in logics, mathematics, computer science, cognitive science, philosophy, art and architecture about intelligence and modeling. Some central concepts and positions are derived from looking into potentially novel architectural modelling approaches. - The first section (‘On Models and Modeling‘) starts this chapter with a more abstract and general aspect of scientific modeling. It identifies two complementary types of models to support progressive thinking. - The second section (‘On Cognition and Memory‘) looks into the virtual capacity for human cognition to produce meaning. - The third section (‘On Tokens and Beliefs‘) derives key ideas of dynamics about informational values from theories of communication and information and through the understanding of vision as a modeling technology. Several permutations are applied for generative architectural modeling and support a more acute way to perform in informational terms. - The fourth and last section (‘Architectural Modeling with Neural Potentials‘) finally aggregates these developments into an architectural framework focusing on architectonics and modeling with the intent to contribute to the ideas of computation, the generative, and articulations in a progressive way. Accordingly, this chapter concludes with a synthetic summary of its central theoretical contribution.\newline\par
The third chapter (‘Experiments in Modeling with Neural Potentials‘) gathers more practical endeavours. It investigates the previously developed ideas into practical experiments encompassing a complementary model of intelligence at the interface between computers and computational accounts of human cognitive vision (in more technical terms, visual ‘Brain-Computer Interfaces‘, or BCI). It is divided into two main sections. - The first section (‘Corollary References‘) serves as an initial review of the state-of-the-art research in this multidisciplinary field. While progressively identifying key ideas and methods beyond clinical and assistive applications for the subsequent development of experiments, it categorises main components from data acquisition to signal processing and decoding towards the scheme of a general research framework. The second section (‘Conducted Experiments‘) applies these findings progressively and tries to derive existing BCI models of natural communication of spelling words towards aggregating parts in the visual mode. It begins with the generalisation of a reference model known as the ‘P300-BCI Word Speller‘ and points to the generalisation of visual discrimination and its methods for visual modeling. It continues with an adapted version of the spelling model into the generation of shapes and investigates the encoded meaningfulness and its potential dynamics. A third and last experiment augments the reflection. It develops a last computational model which enables both the human capacity for visual discrimination in complex environments and the computer capacity for generating large amounts of probable solutions in a vast solution space. The experiment takes the shape of iteratively aggregating non-trivial parts and posits that such a model enables non-syntactical accounts of generative architectural modeling.
Similarly to previous chapters, it concludes with a summary of found methods in encoding design methods and their segmented domains for a potentially richer exploration of generative solutions in future architectural modeling.
The conclusion draws a synthetic outlook of the thesis and its developed arguments. It starts by abstracting a contribution to the idea of ‘intelligent‘ architectural modeling by the way the very modeling of intelligence may show a path to continue with the benefits of computation in architecture and suggests other generative approaches than syntactical ‘a priori‘, in the realm of information and communication. In the positive perspective of pursuing this research, some limitations and near-future developments are mentioned to redefine certain grounds and perspective applications as a new point of departure. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000604304Publication status
publishedExternal links
Search print copy at ETH Library
Publisher
ETH ZurichSubject
Design Computing and Cognition; Architectural Modeling; Generative Design; Neurodesign; Machine Learning; Brain-Computer InterfacesOrganisational unit
03563 - Hovestadt, Ludger / Hovestadt, Ludger
More
Show all metadata
ETH Bibliography
yes
Altmetrics