A multi-disciplinary conversation about Collective intelligence – reflection on some missing elements

The mobile phone and its connectivity with the internet allows growing numbers of us to find, connect, and interact. That was the technological foundation for the original vision of cMOOCs (developed by George Siemens and Stephen Downes) as a democratic learning experience, and connectivism as a learning theory for a digital age (Siemens 2005).

Collectivism, as discussed in a multi disciplinary context at Nesta http://www.nesta.org.uk/event/roots-collective-intelligence, relies on the same technological infrastructure. What is collectivism and how does it differ from connectivism? We heard that the Internet allows for a new form of activism for example, political. How do individuals participate? We were given the example of number of Facebook likes as an indicator of collective intentionality, a starting point. However, how to sustain collectivism i.e. participation, was a recurrent question; how does an individual move on from the Facebook click to participate?  Participating in connectivist learning opportunities requires good infrastructure design and from the individual curiosity, motivation to learn, to contribute, openness, digital literacy (a fair amount) and I would argue that digital fluency and socio-emotional experience are the elements that sustain on-going participation.

Intelligence. Summing up the day Colin Blakemore focused on intelligence. His considered opinion was that we hadn’t spent enough time talking about what we meant by intelligence. He is a cognitive neuroscience and so unsurprisingly, he also spent some time reflecting on the contributions made by the speakers who were cognitive scientists. Those mainly cohered around what, when and how collective intelligence occurs with when and what depending on group size (Robin Dunbar) how on noise reduction, competence, persistence, reputation (Chris Frith). Others concentrated on decision making as investigated in the laboratory; the role of argumentation in decision-making and the quality of decision making. This was a useful contribution as we had already heard that a binary response as the norm could be the problem for collectivism in political contexts. Also, it has some connection with learning science, where design for collaborative learning (i.e. group work) requires a joint task and the opportunity to propose ideas, discuss them with others, negotiate a solution to disparate ideas, and contribute to a shared outcome.

Overall, the multi-disciplinary approach was very successful. However, there was a sense that the individual was objectified. The subjective was largely missing from the day; purposefulness and socio-emotional experience. Therefore, it was interesting to learn about the fall off in Wikipedia contributions since bots were introduced. The explanation provided was that the bots were overly obedient to protocol and rules - another example of the clunkiness of bot interaction as described in a previous post and that the uniquely human nature of socio-emotional experience is so often neglected

Surely, design is important both for successful knowledge building and the socio-emotional experience of collective action.

Siemens, G. (2005). Connectivism: A learning theory for a digital age. International journal of instructional technology and distance learning, 2(1), 3-38.

Spending a day with the Emotion Historians

This event https://emotionsblog.history.qmul.ac.uk/2015/09/tears-and-smiles-programme/ was an opportunity to exercise some transdisciplinary thinking. Could emotion historians contribute to the scholarship of online and distance education?  Specifically, the issue of disembodiment when interpersonal interaction takes place online.

Descriptors assigning purpose to an emotion (e.g. mocking, patronising) peppered the day and emotion as culturally situated performance for control, or to resist control, was a strong theme to emerge. Emotion as a rational act expressed for strategic purposes although, there was some debate on ‘the complex emotion repertoire’ of Margery Kempe (a medieval mystic). While those around Kempe  perceived her behaviour as irrational it could be explained as a manifestation of illness. Self-report by Kempe, that sensory stimuli could  trigger strong emotion, would fit with the experience of some with neural evidence of temporal lobe epilepsy. 

Emotion as experienced and in particular socio-emotional experience, those feelings and thoughts embedded in the dynamic of an interpersonal interaction, did not feature. Understanding social emotions is important for a socio-cultural pedagogy based on the idea that students will co-construct knowledge through discussion, through sharing ideas. The negotiation of ideas can generate strong emotions and when it takes place online it is through written communication and is mediated by a digital device i.e. the physical other is unseen. ‘The textual face of the medieval poet’ did provide an eloquent account of using literary device such as metaphor and the multimodal  (through styling for example, embolden) to express emotion and achieve emphasis through writing  (as is the case when people interact online).  But once again, this account was confined to the expression of emotion and one-way interaction i.e. poet to reader.

Art history provided considerable resource throughout the day, an observer interpretation of an expressed emotion that is then reinterpreted by the historian.    By contrast social emotions rely on language,  spoken or written, and their history would require a different resource. So, while I enjoyed an illuminating set of talks, thank you, I am still searching for the history of those emotions, experienced and expressed, that are embedded in the social.

What would be the point of Education if computers replace teachers?

Turing’s work on developing early versions of computers led to the imitation game as a way of investigating whether computers can think? The imitation game (Turing test) involves a participant in one room, a human confederate (someone whose behavior is under the direction of the experimenter) in another room and a computer terminal containing a program that simulates intelligence also in another room. The task for the participant is to decide which is human. In 1951 Turing predicted that by 2000 the average person, naïve to the fact that an interaction partner was a computer program, would assume humanness 70% of the time. Undoubtedly these ideas inspired the foundation of Artificial Intelligence (AI). Its subsequent evolution is brilliantly summarized on the BBC’s iWonder website http://www.bbc.co.uk/timelines/zq376fr.  In July 2015 an article on research that uses an echoborg (a human who acts as the mouthpiece for an artificial intelligence system) for the Turing test was featured on the BBC Futures website http://www.bbc.com/future/story/20150717-the-people-possessed-by-computers and for one week in September 2015  AI was a featured topic on the BBC reflecting a contemporary fascination with sharing our personal and working spaces with intelligent agents

However, the penetration of AI into Education (AIED) was not covered. AI techniques mean that artificial agents could replace teachers. An Intelligent tutor (IT), based on adaptive AI systems, means that learning contexts can be personalized for a student. The system (IT) uses a teacher-pupil model to adjust the learning task to an appropriate level and a task model to provide appropriate feedback to the student. Pedagogically sound opportunities to learn can be extended in both reach and frequency. A teacherbot has been used to answer student questions on a massive open online course (a MOOC). Teaching presence on MOOCs is sparse so that students who enroll on a MOOC need not incur a financial cost. Therefore, researchers are investigating ways in which the teacher can be assisted, or replaced by, a bot. https://www.timeshighereducation.co.uk/news/ask-teacherbot-are-robots-the-answer/2020326.article. Other examples of AI contributions include the ECHOES project that targets atypical development issues by using virtual agents to engage with children on the autism spectrum in order to improve their communication skills. http://echoes2.org/?q=node/2.

Versions of the headline ‘Intelligent agents replace teachers’ are increasingly common. ‘What if’ this transpires? with teachers replaced by AI products (robots, bots, virtual agents). What questions does it raise for Education? For example, what would it mean for a socio-cultural pedagogy that AI products are linguistically challenged and on current appraisal likely to remain so? Furthermore, the interdependence of emotion and cognition when learning is increasingly recognised, with empathy a foundational element of successful interpersonal interaction.  There is evidence from Neuroscience that both language and empathy are uniquely human capacities with dedicated brain structures and neural pathways. For a socio-cultural perspective meanings arise through social interaction and enable us to pursue personal goals and to think beyond the actual. It is social interaction that enables creativity. Therefore, it is significant that a recent report from Nesta concluded that ‘creative occupations are more future-proof to technologies like machine learning and mobile robotics’ http://www.nesta.org.uk/publications/creativity-vs-robots.

That the ability to transform knowledge and create history through social interaction is a uniquely human ability can be supported by evidence from Neuroscience. However, according to Kevin Warwick http://www.kevinwarwick.com, by comparison with computers, our human ability to communicate information is slow, limited in reach, and error prone and our reliance on language may become ‘excess baggage’. Educational Neuroscience, a multidisciplinary area that assesses the potential of Neuroscience for informing Education, could usefully contribute to the debate.

Wearing cat’s ears; you may amuse but don’t be misled by simplistic accounts of Neuroscience

Unsurprisingly, an image of Nicola Sturgeon (First minister for Scotland) wearing a pair of tartan cat’s ears at the launch of a hi-tech digital school was broadcast through both print and social media http://www.dailymail.co.uk/news/article-3222275/What-thinking-Sturgeon-tries-bizarre-brain-scanner-enormous-TARTAN-ears.html.

When used as a form of entertainment (or publicity) there is no harm done. However, this device is marketed as a way of augmenting the human body when communicating mood http://www.necomimi.com/WatchTheVideo.aspx and it has been suggested that it could be used by teachers to assess attention in class in real time; that the ears will ‘prick up’ when the wearer is paying attention. According to the instructions one sensor should be placed above the eyebrow and the other clipped to the ear so that the forehead sensor can ‘read’ the electrical impulses generated by neurons firing in the brain. It is claimed that by using this data to control the motor that positions the cat’s ears they can reflect your mental state.

Do the scientific claims for this device hold up? In medical contexts the EEG has been used as an investigative procedure for over 60 years with technical innovations such as solid-state amplifiers and digital methods of analysis being incorporated as they became available. The collection of the EEG record, and its subsequent interpretation, requires expertise based on years of training. You would be unlikely to find a practitioner who would support the claims made for this device. There is a strong likelihood that the electrical impulses that drive the motor are myogenic (originate in muscle) rather than neurally based. When the biosensors are placed as recommended the one above the eyebrow will pick up both eyeblinks and activity in the frontalis muscle (the muscle that you use to raise your eyebrows) while in noisy environments the ear clip will be susceptible to activity in the post auricular muscle as described here http://brain.oxfordjournals.org/content/100/1/19?ijkey=ecce332eef28da3f167b1373941635f7b915e274&keytype2=tf_ipsecsha and movements of the wearer’s own ears that are under voluntary control.

A single channel electro encephalogram (EEG) is the technical description for this device. There are other manufacturers of such devices which are marketed with suggestions for their use that range from managing a brain-training program to meditation. By contrast a neurophysiologist would expect to have access to at least 21 channels of EEG recorded concurrently from at least 21 biosensors positioned to sample electrical activity over the whole skull area. The neurophysiologist’s skill lies in the interpretation of the patterns of activity across all biosensors. Although the presence and/or amplitude of the alpha frequency is a valid indicator of alertness it is optimally recorded with biosensors placed at the back of the skull. Feel for the midline boney projection at the back of your skull, 5 cms above that would be a good placement. Encouraging a class to use such a device to critically investigate the underlying Neuroscience would have value. Suggesting to teachers and students that such devices are a valid method for monitoring attention and mood is misleading.

Teacherbot, useful tool or just another example of the allure of innovation

When I read the following article

Teacherbot: interventions in automated teaching by Sian Bayne. Teaching in Higher Education. Published online 16 Apr 2015. http://www.tandfonline.com/doi/10.1080/13562517.2015.1020783

I noted with interest that students had dropped the descriptor ‘teacher’ from the bot developed for the E-learning and digital cultures MOOC (University of Edinburgh).  Perhaps like me students could not discern any teacherly function of botty (the name that students adopted for teacherbot) that could not be otherwise achieved by providing a repository of curated resources and a frequently answered questions (FAQ) list.

Undoubtedly the agent that follows the MOOC hashtag (#edcmooc) and harvests student questions and comments is useful, but what did botty achieve? The behaviour of teacherbot (a rule based agent developed by the teachers to automate tweet responses) is problematic for me. Instead of making queries about the course (for example, about assessment procedure) available to all students in their collated form i.e. as a list of FAQs, botty responds to each student individually through twitter. Furthermore, although 60% of the enrolled students are postgraduates (as is the case for many MOOCs) with a professional interest in digital education botty pushes individual resources to individual students. Surely, for this demographic it is access to a curated resource repository that is valuable. 

Teacherbot is positioned as a research initiative that seeks new ways of understanding and developing automated teaching through adopting a critical posthumanist perspective.  While arguing for a critical posthumanist approach Bayne reviewed a previous initiative for automating teaching that was funded by the ESRC http://www.tel.ac.uk highlighting efficiency and productivity as the dominant discourse for that program. However, that section of the Bayne article mystified me. A closer look at the title of one of the targeted articles[1] offers a very different interpretation; higher quality and more effective learning are the outcomes that were envisioned. Nevertheless, Bayne develops a narrative that characterises the current conversation about automated teaching as dispute, with efficiency gains and teacher resistance oppositional views that need to be resolved. By response teacherbot is positioned as  ‘an assemblage of the human and non-human’ unhindered by humanistic assumption and efficiency agendas and thereby provides a new way of revisiting teacher automation.

Does teacherbot make a research contribution? The article is eloquently written and intriguing to read. However, the outcomes are mainly descriptive and anecdotal. As a teacher (and on occasion student) continuously and actively involved in online teaching and learning for more than two decades I have never considered the automation of teaching to be a threat.  Indeed, I can envisage how automation (for example, the collection and collation of learning analytics (LA) data) has promise and I would not object to it being considered as co-teaching provided that the automated information and myself were appropriately assembled so that LA information is monitored throughout.  However, I cannot discern any evidence to suggest that botty and the MOOC teaching team are entangled in a way that merits a claim of co-teaching or that this initiative has informed the scholarship of online teaching in any useful way. Furthermore,  the evidence presents botty as not very accomplished when it comes to recognising and responding to the socio-emotional needs of students – perhaps because it is not human.

1.         Laurillard, D., Productivity: Achieving higher quality and more effective learning in affrodable and acceptable ways. 2011.

http://www.tlrp.org/docs/ProdBeta.pdf.

Taylor & Francis Group partners with researchers, scholarly societies, universities and libraries worldwide to bring knowledge to life.  As one of the world’s leading publishers of scholarly journals, books, ebooks and reference works our content spans all areas of Humanities, Social Sciences, Behavioural Sciences, Science, and Technology and Medicine.   From our network of offices in Oxford, New York, Philadelphia, Boca Raton, Boston, Melbourne, Singapore, Beijing, Tokyo, Stockholm, New Delhi and Johannesburg, Taylor & Francis staff provide local expertise and support to our editors, societies and authors and tailored, efficient customer service to our library colleagues.