Centre for Cognitive Science (COGS)

Autumn term 2013

• Oct 1st Siddharth Narayanaswamy (Purdue): Compositionality in vision and language

Compositionality can be found almost everywhere one looks. It is manifest in as diverse a range of entities as objects around us, the languages we use, and even our actions and interactions with the world. My work involves exploring and exploiting the general nature of such compositionality, often across multiple modalities, to solve deep and complex problems in perception.

I demonstrate such ability in a variety of domains including part-based structures, board games, and activity recognition. I also show evidence for a particular kind of compositionality in how the brain perceives the world, lending further credence to the ubiquity and utility of compositionality.

• Oct 8th Paul Verschure (Pompeu Fabra & Catalan Institute of Advanced Studies): The Distributed Adaptive Control theory of mind, brain and behavior

The brain evolved to maintain a dynamic equilibrium between an organism and its environment. We can define the fundamental questions that such a brain has to solve in order to deal with the how of action in a physical world as: why (motivation), what (objects), where (space), when (time). I call this the H4W problem. Post the Cambrian explosion a second factor became of great importance for survival: who and now brains adapted to the H5W challenge.

I will present the hypothesis that consciousness evolved to enhance fitness in the face of H5W. My theory of mind and brain, or Distributed Adaptive Control (DAC)  shows how H5W can be solved through the interactions across multiple layers of neuronal organization and assigns a specific role of consciousness in the optimization of the real-time control of action. DAC makes specific predictions on both the structure and function of the neuronal correlate of consciousness that I will discuss with respect to memory, decision making and attentional processing.

• Oct 15th Phillip Prager (Cambridge): The cognitive science of Duchamp and Dada

Dada is the infant terrible of art history, an anarchic movement that is typically referred to as nihilistic, pathological, and firmly enshrined within the modernist paradigm and the context of WWI. Through the lens of classical, romantic, and psychoanalytic notions, it certainly appears almost antithetical to creativity. Yet from a cognitive point-of-view, Dada marks a watershed in the understanding of creativity, and articulates principlesof creative cognition with surprising insight and precision many decades ahead o science.

• Oct 22nd Matthias Scheutz (Tufts): Moral robots? Investigating human perceptions of moral robot behaviour

Autonomous social robots are envisioned to be deployed in many different application domains, from health and elder care settings, to combat robots on the battlefield.  Critically, these robots will have to have the capability to make decisions on their own to varying degrees, decisions that should ideally conform to moral principles and ethical expectations.  While there is currently no autonomous "explicit ethical agent" (in Moore's sense) that can make truly informed ethical decisions, we can still start to empirically investigate human reactions to such agents as if they existed using "Wizard-of-Oz"-style experiments.  For it is important to understand not only how to make robots identify and properly process ethical information, but also how such robots would be perceived and accepted by humans.  In this presentation, I will provide an overview of our recent work on evaluating human perceptions of and reactions to "moral robots" in human-robot interaction experiments.  Specifically, I will report results from studies in which human subjects were exposed to different types of robotic agents (with human-like and robotic appearance) that may protest human commands for a range of (ethically justified) reasons. Critically, the results suggest that humans are inclined to take robot protest seriously and that robot appearance does not seem to have an influence on human judgments.

This talk is a joint COGS seminar/satellite talk of the EUCOG members meeting on Social and Ethical Aspects of Cognitive Systems.


• Oct 29th Peter Cheng (Sussex): Explorations in knowledge re-codification (Part 2): On the design of novel notational systems for propositional logic (and syllogisms)

It has been well and long-established in cognitive science that the external representations we possess substantially determine how we reason and solve problems, with task demands varying by (over) an order of magnitude under alternative representations.  This talk applies the principles gleaned from my previous designs of novel representations to the creation of novel notational systems for propositional (Boolean) logic (and syllogisms).  Unlike conventional notations for these topics, which use just sentential or diagrammatic means to encode domain information, the new representations — Truth Diagrams (and Category Pattern Diagrams) — combine both linear concatenation and graphical schemes to integrate the core concepts, on multiple levels, that are inherent in these topics.

• Nov 5th NO SEMINAR - POSTPONED UNTIL SPRING TERM:  Blay Whitby (Sussex): The Milgram experiment – An historical appraisal

• Nov 12th CHANGE OF SPEAKER: Blay Whitby (Sussex): Flying Lessons: What aviation tells us about cognition in the real world

• Nov 19th Bill Fulford (Oxford): Delusions and Spiritual Experience: Case Studies and a COGS Research Challenge

• Nov 26th Dave Cliff (Bristol): Robot traders in the financial markets: WTF?

Until about 10 years ago, most traders in the world's financial markets were highly-paid humans. In the last decade, these humans have been increasingly replaced by adaptive automated trading systems, known in the business as "robot traders". Robot traders perform at literally super-human reaction-speeds and data-processing bandwidths, and threaten to render human traders largely redundant. As this new and risky technology has become more common in the markets, we are seeing more evidence that there may be new and worringly major systemic risks; direct consequences of unforseen interactions among robot trader systems. This, coupled with ineffective regulation and weak-willed politicians, means that a major collapse in the financal markets could potentially occur for reasons of technological, rather than economic, failure.

• Dec 3rd Yorick Wilks (Oxford): What will a companionable computational agent be like?

The talk begins by looking at the state of the art in modeling realistic conversation with computers over the last 40 years. I then move on to ask what we would want in a long-term conversational agent that was designed for a long-term relationship with a user, rather than the carrying out of a single brief task, like buying a railway ticket. Such an agent I shall call “companionable”: I shall distinguish several functions for such agents, but the feature they share will be that, in some definable sense, a computer Companion knows a great deal about its owner and can use that information..

By way of illustration, the lecture describes the functionality and system modules of a Senior Companion (SC), one of two initial prototypes built in the first two years of the EC Companions project. The SC provides a multimodal interface for eliciting and retrieving personal information from the elderly user through a conversation about their photographs. The Companion, through conversation, elicits life memories, often prompted by discussion of their photographs. The demonstration is primitive but plausible and one of its key features is an ability to break out of the standard AI constraint on very limit pre-programmed knowledge worlds into a wider, unbounded world of knowledge in the Internet by capturing web knowledge in real time, again by Information Extraction methods. The lecture finally discusses the prospects for machine learning in the conversational modeling field and progress to date on incorporating notions of emotion into AI systems.

• Dec 10th Angelo Cangelosi (Plymouth): Embodied language learning: From sensorimotor intelligence to symbols

Growing theoretical and experimental research on action and language processing and on number learning and space representation clearly demonstrates the role of embodiment in cognition and language processing. In psychology and neuroscience this evidence constitutes the basis of embodied cognition, also known as grounded cognition (Pezzulo et al. 2012). In robotics, these studies have important implications for the design of linguistic capabilities in cognitive agents and robots for human-robot communication, and have led to the new interdisciplinary approach of Developmental Robotics (Cangelosi & Schlesinger 2014). During the talk we will present examples of developmental robotics models and results from iCub experiments on the embodiment biases in early word acquisition studies, on word order cues for lexical development and number and space interaction effects. The presentation will also discuss the implications for the “symbol grounding problem” (Cangelosi, 2012) and how embodied robots can help addressing the issue of embodied cognition and the grounding of symbol manipulation use on sensorimotor intelligence.