W.F.G. Haselager

Connectionist semantic systematicity

Fodor and Pylyshyn [Fodor, J. A., & Pylyshyn, Z. W. (1988). Connectionism and cognitive architecture: A critical analysis. Cognition, 28, 3-71] argue that connectionist models are not able to display systematicity other than by implementing a classical symbol system. This claim entails that connectionism cannot compete with the classical approach as an alternative architectural framework for human cognition. We present a connectionist model of sentence comprehension that does not implement a symbol system yet behaves systematically. It consists in a recurrent neural network that maps sentences describing situations in a microworld, onto representations of these situations. After being trained on particular sentence-situation pairs, the model can comprehend new sentences, even if these describe new situations. We argue that this systematicity arises robustly and in a psychologically plausible manner because it depends on structure inherent in the world.

Stimulating the Self: The Influence of Conceptual Frameworks on Reactions to Deep Brain Stimulation

Deep brain stimulation (DBS) is generally considered to have great practical potential. Yet along with its remarkable efficacy, which is currently being tested in application to many pathologies, come a certain number of complications. In particular, there seem to be several adverse psychological implications behind a relevant number of postoperative situations. Whether these effects have to be attributed to a reactive response to a new situation or whether they are caused by the stimulation itself, or both, remains to be elucidated. This article takes up this issue, and evaluates how various views and conceptual schemes concerning the mind–brain relationship might play a role in the ensuing (mal)adaptation to DBS treatment. It is valuable to investigate the relationship between different conceptual assumptions and the onset of psychosocial maladaptations in clinical settings. In particular, we hypothesize that the frequently reported maladaptations might be partially caused by a concep-tual shift away from dualism and toward a “neurocentric” materialism, promoted by the scientific explanation of the pathological situation. This article examines which conceptual framework is likely to play a dominant role in the perception of the general public of how the mind relates to the brain, and how this contrasts with the opinions of experts in cognitive neuroscience. We investigate how these different conceptual frameworks might play a role in the occurrence of psychosocial maladaptations to DBS treatment in clinical settings. Finally, we suggest how a more inclusive, embodied embedded stance toward the mind–brain relationship might help to overcome such maladaptations.

Connectionism, Systematicity, and the Frame Problem

This paper investigates connectionisms potential to solve the frame problem. The frame problem arises in the context of modelling the human ability to see the relevant consequences of events in a situation. It has been claimed to be unsolvable for classical cognitive science, but easily manageable for connectionism. We will focus on a representational approach to the frame problem which advocates the use of intrinsic representations. We argue that although connectionisms distributed representations may look promising from this perspective, doubts can be raised about the potential of distributed representations to allow large amounts of complexly structured information to be adequately encoded and processed. It is questionable whether connectionist models that are claimed to effectively represent structured information can be scaled up to a realistic extent. We conclude that the frame problem provides a difficulty to connectionism that is no less serious than the obstacle it constitutes for classical cognitive science.

Creativity: Surprise and abductive reasoning

This paper investigates creativity focusing on the nature of abductive reasoning, as originally formulated by Peirce, situating it in the context of the theory of self -organization. An ancient question will be addressed: is it appropriate to investigate creative processes from a mechanistic perspective or do they involve subjective elements which cannot - in principle - be investigated from a mechanistic view? This question will guide our investigation, which has as an initial hypothesis that creativity starts with surprise and involves a self -organizing process in which abductive reasoning occurs allowing the expansion of well-structured set of beliefs. This process is considered a part of the establishment of habits in self -organizing systems. We argue that a deeper understanding of how self-organizing processes involving abductive reasoning may take place in creative systems could elucidate the complex debate about the mechanical versus nonmechanical ingredients of creativity.

Towards a computational model of the self-attribution of agency

In this paper, a first step towards a computational model of the self-attribution of agency is presented, based on Wegners theory of apparent mental causation. A model to compute a feeling of doing based on first-order Bayesian network theory is introduced that incorporates the main contributing factors to the formation of such a feeling. The main contribution of this paper is the presentation of a formal and precise model that can be used to further test Wegners theory against quantitative experimental data.

Goals are not implied by actions, but inferred from actions and contexts

People cannot understand intentions behind observed actions by direct simulation, because goal inference is highly context dependent. Context dependency is a major source of computational intractability in traditional information-processing models. An embodied embedded view of cognition may be able to overcome this problem, but then the problem needs recognition and explication within the context of the new, layered cognitive architecture.

BCI and a User’s Judgment of Agency

Performing an action with the assistance of a BCI may affect a user’s judgment of agency, resulting in an illusion of control, or automatism. We analyze this possibility from a theoretical perspective and discuss various factors that might influence a user’s judgment of agency in a BCI context. We present two pilot experiments that illustrate how this theoretical possibility can be investigated experimentally. We examine potential psychological, ethical, and legal implications of mistaken judgments, and potential benefits of the constructive manipulation of a user’s judgment of agency.

Human SLAM, Indoor Localisation of Devices and Users

The indoor localisation problem is more complex than just finding whereabouts of users. Finding positions of users relative to the devices of a smart space is even more important. Unfortunately, configuring such systems manually is a tedious process, requires expert knowledge, and is sensitive to changes in the environment. Moreover, many existing solutions do not take user privacy into account. We propose a new system, called Simultaneous Localisation and Configuration (SLAC), to address the problem of locating devices and users relative to those devices, and combine this problem into a single estimation problem. The SLAC algorithm, based on FastSLAM, is able to locate devices using the received signal strength indicator (RSSI) of devices and motion data from users. Simulations have been used to show the performance in a controlled environment and the effect of the amount of RSSI updates on the localisation error. Live tests in non-trivial environments showed that we can achieve room level accuracy and that the localisation can be performed in real time. This is all done locally, i.e. running on a users device, with respect for privacy and without using any prior information of the environment or device locations. Although promising, more work is required to increase accuracy in larger environments and to make the algorithm more robust for environment noise caused by walls and other objects. Existing techniques, e.g. map fusing, can alleviate these problems.

Useful review marked by conceptual vagueness [Open Peer Commentary]

To this day, the experimental design of Libet and colleagues (1983) is commonly applied in neuroscience research on intended action. Therefore, the review by Saigle, Dubljevic, and Racine (2018) is most welcome as it provides a useful tool for checking the consistency across the various results of these Libet-type studies. In addition, the authors construct a taxonomy of Libet-type experiments (see their Table 2 and Supplementary Tables 1 and 2), which provides a useful overview of existing variations within the Libet design. Moreover, their review highlights the inconsistencies within and between publications: Often only averages across participants are reported, no standard deviations are provided, or certain measurements are recorded but not reported. These findings emphasize the importance of concise research reports that allow valid and complete comparisons between related studies. The main goal of the authors is to reflect on what Libettype studies can tell us about the complex phenomenon of free will. A review of the scale they deliver provides an excellent opportunity to detect the core results that are found consistently across studies and are in some way relevant to our understanding of free will. However, in our view, the authors failed to fully seize this opportunity. First, they did not provide their definition of free will. Without such a foundation, the discussion is marked by conceptual vagueness. Second, the authors seem to take for granted that Libet studies can tell us something about the neuromechanism of free will. However, why such a mechanism would exist and what it could look like remain unspecified, which makes it harder to see exactly why and in what way Libet-type studies are relevant to our understanding of free will.

Hierarchical structure in prefrontal cortex improves performance at abstract tasks

information and is set to either low or high activity. This abstract information corresponds to the context in a simple version of WCS: either the cards have to be sorted according to color or according to shape. Depending on this context (or current policy), either the first half of the rest of the input nodes is relevant to determine the desired output, or the second half. These input nodes thus represent the color and shape in WCS, respectively. The input nodes that are not relevant in the current trial are distractors. The task is illustrated in Fig. 2. The experimental condition in this task is the number of low-level input nodes. If our hypothesis is correct, preserving the information within the abstraction levels is most important with a large number of low-level input nodes, as the abstraction is most important in this situation. On the other hand, integrating information between the abstraction levels is most important in the conditions with a small number of low-level input Fig. 2. The version of the policy abstraction task with three input nodes (and thus two low abstraction input nodes). Each column represents one trial. The input pattern at the top is used to set high and low activities in the input nodes at the beginning of each trial. In the lookup table the top row represents the abstract input that is used to determine whether the middle row or bottom row is currently important to find the desired output. The x in the lookup table is a wildcard: the input node can be either 1 or 0, it is irrelevant for the task and thus a distractor. The letters in the output row correspond to the four output nodes in this task, with the highlighted letter corresponding to the output node that should show the highest activation at the end of the current trial. February 1, 2013 12:9 WSPC Proceedings Trim Size: 9in x 6in article