Two Cultures
Independent film-maker, Holly Stead, has been following the NOBELini project since its launch in May 2009 helped by sound recordist, Tim Bamber. Following the exhibition in February 2010, she hopes to create a full-length documentary about the project.
Here we share her insights into the project so far…
A networking event followed the NOBELini: Blind Data private view on Thursday 11th February 2010 inviting scientists and designers to seed ideas for science-design projects.
(Photos by Cléon Daniels)
After watching Holly Stead’s short film edit, Two Cultures, which documents the NOBELini project, participants each wrote down 4 words on a post-it note to reflect their interests. Designers and scientists took it in turn to choose post-it notes from the board and find their owners. Brainstorming, if fruitful, seeded ideas, which were documented on SEED cards. Finally, SEED cards were displayed and participants voted (with heart post-it notes) for their favourite idea(s).
Based on the responses we received, 93% of participants enjoyed SEED dating. 87% found it personally useful and 47% professionally useful. Comments in this regard include:
“Really refreshing, enabled me to think about my work in a different way”
“Always good to look beyond your own subject”
73% of respondents were happy with the way the event was organised. Comments for ways to improve the event include:
“Short intros from everyone then general melée”
“More time to chat”
“Clearer instructions, fewer instructions at a time”
“Publicise more, more information on website, ask people to send forth ideas”
87% of respondents said they would like to participate again in a similar event and 53% said they would follow-up on the new contacts they had made.
Words that scientists and designers used to describe their interests include:
“Narrative”
“Metamythemagic”
“Communication”
“Sustainable”
“Vision”
“Dancing”
“Absurdity”
“Pattern”
“Worms on prozac”
“Brain”
“Nature”
“Neuroscience”
“Cells”
“Knowledge”
“Life”
The question: What would you design if anything were scientifically possible?
SEEDs and their love-heart scores are tabulated below:
| By (scientist and designer) | SEED idea | Love heart score |
| (Lieven and Fernanda) | “A biological external hard-drive to access lost memory in patients with Alzheimer’s.” | 8 |
| (Nandita) | “The tangible self: what if we could decode our daydreams? If we could map out individual thoughts and innermost desires we could simplify decision-making.” | 7 |
| (Shirin and Stacey) | “Use of scent in memory loss to target memory neurons” | 4 |
| (Fauzia and Fernanda) | “Make all scientific theories visually accessible: tactile and/or sensory” | 3 |
| (Natalie and Heather) | “White blood cell model for disaster relief” | 3 |
| Brenda and Fernanda) | “Everything that is destroyed becomes something new and useful?” | 3 |
| (Brenda and Heather) | “Personal algal growth pods for energy” | 3 |
| (Stephen and Jana) | “A human brain composed of hundreds of worms” | 2 |
| (Fauzia and Heather) | “Crystal cities inspired by microscopic structures” | 2 |
| (Shirin and Stacey) | “A wearable record of life/memory/experience” | 1 |
| (Stefan and Stephanie) | “Self assembling seating in old people’s homes that changes configuration every day. Each seat recognizes the individual and their medical/emotional needs.” | 1 |
| (Shirin and Olivia) | “Remote-controlled cell-growth” | 1 |
| (Emily and Andree) | “An instant interactive electronic forum for dissemination information about discovery—signaling advances across disciplines” | 1/2 |
| (Anon) | “A quantum computer to predict the lottery results” | 1/2 |
| (Anon) | “Recycle our CO2 for plant/algal consumption” | 0 |
| (Jenny and Jana) | “Aid for blind people that projects a camera/computer-rendered image directly into the brain” | 0 |
| (Brenda and Stephanie) | “Natural systems for man-made jobs” | 0 |
| (John and Fernanda) | “A way to visualize and understand fantasies” | 0 |
| (Stefan and David) | “Bacterially-driven power generation” | 0 |
| (Stefan and Kostya) | “Retune wavelengths to increase energy output for power generation” | 0 |
Final comments from respondents included:
“Missed out being on email list of participants”
“The chance factor is most interesting because it creates new challenges”
“Really interesting people, designers receptive and everyone I met very interesting”
“Any non-scientist has been exposed to so many different ideas, it must inspire them”
The nationwide NOBELini scheme welcomed applications from a broad spectrum of designers and scientists. 60 applicants (30 science, 30 design) were shortlisted to participate in a speed-dating event at the Science Museum’s Dana Centre in May 2009. Participants specified collaborative preferences and 30 scientist-designer pairs were devised.
Pairs were given until September 2009 to brainstorm and devise proposals for designs, which celebrate and/or communicate science across 4 themes: stem cells, energy and recycling, synthetic and systems biology and imaging. A total of 24 proposals were subsequently reviewed and graded by an international jury of professional designers and scientists.
The three most highly scoring proposals were awarded prizes of £2000 to develop their designs. Design prototypes were exhibited at the Dana Centre in February 2010. The exhibition launched on Wednesday 10th February 2010 with a speech from Sir Tim Hunt, who participated in the sibling project, Nobel Textiles and Professor Amanda Fisher (Director, MRC Clinical Sciences Centre), Fabrics of Life project pioneer.
(Photos by David Nelson)
Evaluation and Feedback
Public feedback was collated from responses provided by exhibition visitors. 100% of respondents enjoyed the exhibition and 80% said they were stimulated to learn about the science behind the design work. Half of the respondents specifically requested to be kept informed of future developments and/or projects of this nature.
What did visitors like?
“Presence of collaborators to explain work”
“Layout and explanation”
“Exploration of scientific concepts in a new light”
“Concept, applications”
“Interactivity”
“Integration of imagination”
“The way science-design is developing”
“Laid back atmosphere, good event networking with scientists/artists”
What did visitors not like?
“Lengthy explanations; short summary better”
“More lay-friendly explanations”
“Exhibits needed more obvious explanation”
“Lack of info on the website”
“Too small a place”
“Too hot”
As part of the exhibition and playing on the Valentine theme, organisers Carole Collet (Central Saint Martins College of Art & Design) and Brona McVittie (MRC Clinical Sciences Centre) initiated the first in a series of SEED-dating events to bring scientists and artists together. Our first event on Thursday 11th February 2010 at the Dana Centre explored what designers would create were anything scientifically possible. Read more here…
We often live life reflexively, unaware of our mental state, emotions, and behavioural patterns. What are we feeling? How do others perceive us? And, can we push ourselves to break through involuntary behavioural patterns and achieve our own self-dictated moods?
(Photo by Cléon Daniels)
Feel Out Loud! is an experience that challenges the public with these introspective questions. Visitors’ moods are continuously captured from facial expressions. These moods are graphically reinterpreted as an abstraction of dense networks, subtly reminiscent of their origin in neurons and brain. Experiments have shown that by simply changing our outward emotional display (for example, by simulating a smile) we can change our internal state of feeling and move it towards that emotion. The installation allows the public to become aware of their emotions and question how their behaviour appears to other people. The deep interplay of feedback loops as visitors become aware of their internal feelings actively pushes them towards the mood of their choice. Feel Out Loud! is a playground for one of the most core human experiences: mental and emotional state.
(Photos by David Nelson)
The general public often questions the large sums of money devoted to scientific research and struggles to understand how results in basic sciences, like biology, relate to everyday life. Feel Out Loud! is the result of a collaboration between Céline Marcq (designer) and Ev Yemini (scientist), whose shared desire to express their research and expertise responds to the need to show the public how new research touches our lives.
Céline (Above left) specialises in textiles that incorporate elements of interactivity and sensory design. Currently at Central Saint Martins College of Art & Design, she explores how patterns come to life through coding and poses questions about how technology might be used to demonstrate development.
Ev (Above right) researches how neural codes translate into behaviour at Cambridge University. He works with the nematode C. elegans, a tiny worm that has been the subject of many Nobel Prizes. Ev is creating an automated, high-throughput system to film these worms and place the analysis into a searchable database for use by scientists investigating how genes and environment influence behaviour.
Science behind the design
Ev Yemini explains the inspiration for Feel Out Loud!
Feel Out Loud! uses a camera and computer to read faces, interpret their mood, and transform these moods into abstract graphical representations. The science behind the design, at a high level, is quite simple. But, the details of recognizing mood from faces rely heavily on complex mathematics and can even be obtuse to researchers considered experts in this field.
The field of facial computer vision dates back to the 1980s and is now in its adolescence. These days we have seen an explosion of products featuring this technology. Digital cameras employ face localization to determine correct focusing for portraits as well as facial expression recognition to capture smiling, non-blinking faces. Security software uses biometric face recognition to spot known criminals, facial expression recognition to discover suspicious characters, and face tracking to follow its targets. And, recent social software uses facial expression recognition to transform video game player’s expressions onto their in-game avatar, develop robots that respond to people’s moods, and help those with autism to better understand the emotions of other people during their interactions.
In our case, we require both face detection to find a face and facial expression recognition to determine the mood expressed on that face. How is face detection done? Currently, the most popular algorithms derive from the one published by Viola & Jones in 2001. If you had to develop a face detection algorithm, you would likely look for eyes, a nose, a mouth, and other salient features. Such algorithms tend to be slow and have difficulty with different skin tones, lighting conditions, and occlusions (e.g. glasses, a beard, and non-frontal faces). Viola & Jones took a different, more raw approach. Their algorithm has 3 steps. First, they transform pictures into a black and white approximation that covers all scales (from small faces to large ones). Second, they use a training set of images, with and without faces, to determine which simple patterns (i.e., 3-4 black and white squares arranged adjacently) are correlated with faces. And third, they build a cascaded set of rules that, based on these simple patterns, quickly cover an entire picture looking for regions with patterns that highly-correlate to faces. Surprisingly, this method is extremely fast and accurate.
Now that we have detected a face, how do we determine its mood? Keep in mind that, in the 1970s, psychologists found strong evidence for 6 universal facial expressions: anger, disgust, fear, happiness, sadness, and surprise. In short, regardless of the culture of those expressing the emotion and those viewing it, people show a consensus of opinion in identifying these 6 expressions. Unlike face detection, algorithms for facial expression recognition have no clear leader amongst them. The best of the best, however, fall into 5 categories that are often weaved together to improve results. All 5 tend to employ transformations that simplify the input picture into a smaller subset of descriptions (descriptions that, while great for algorithmic purposes, are often not very meaningful to human beings). First off are the ones that use similar techniques to the Viola & Jones algorithm mentioned above. Second, are a large group of algorithms that, using the aforementioned subset of descriptions, determine whether there is sufficient statistical evidence for the presence of any of the 6 universal facial expressions. The third group, represent the face in alternate coordinate spaces and check whether it matches any templates for known expressions. Fourth are neural networks; these are trained on large databases full of facial expressions, they then attempt to classify new faces based on their training. And fifth are algorithms that decide which muscles must be active to warp a face into its expression. The muscles activated during each facial expression are well known, matching them back to their representative mood is a trivial task.
Feel Out Loud! uses Visual Recognition’s eMotion software as its underlying engine to determine mood from faces. This software was developed by Professor Theo Gevers in the ISLA lab at the University of Amsterdam. Further information is available here, http://www.visual-recognition.nl.
(Photo by David Nelson)
Science and design have much in common. Both require imagination, innovation and dedication. However, their methodologies differ somewhat. A scientific failure might be a design success. What if design principles were applied to science?
(Photo by David Nelson)
Science is structured and clearly defined; researchers begin with a hypothesis and devise experiments to test the theory. Although unbiased in its process, the scientist hopes to support their hypothesis with positive data. All too often experiments don’t work out that way and the result is termed ‘negative data’. The lucky or clever scientist might later find such negative data to prompt a major discovery, although the more frequent end result is nothing but frustration. Avenues of investigation may thus be dropped in favour of experiments that are ‘working’.
(Photo by David Nelson)
Design is not so much about right or wrong answers but questioning the way we interact with the world. Berit Greinke (designer) and Jay Stone (scientist) have been inspired by their different perspectives to see if design could teach science a ‘new trick’.
The Good, the Bad and the Negative celebrates laboratory disaster. A maze table invites the public to explore hidden successes in apparent cul-de-sacs. And ‘living’ textiles, created from a series of experiments probing camel hair as a substrate for bacterial growth, question the nature of success and failure.
(Photo by David Nelson)
Berit (Above left) is a textile and surface designer whose work incorporates sound using electronics. A recent MA student at Central Saint Martins College of Art & Design, she is currently investigating patterns shared by our senses of sight and of hearing, for example rhythm and composition and how these senses interact to influence our perception.
Jay (Above right) researches how certain genes affect blood flow to the eye. Based at the Institute of Ophthalmology and funded by the MRC Laboratory for Molecular Cell Biology at UCL, she is also a keen communicator and writes for several science websites as part of her desire to share science with the wider public.
Science behind the design
By Jay Stone
I am a second year cell biology PhD student. My research group is based at the Institute of Ophthalmology in London and together we are working on characterising the molecular mechanisms behind an eye disease called Macular Telangeictasia (MacTel).
Our sight affects our interpretation of the world and the way we communicate with those around us. At the centre of our retina we have a specialised region known as the macular, which has no blood vessels. This area of our eyes is home to colour-perceiving cells known as cones. Our cones afford us a high ‘central’ visual acuity: clearness of vision. Patients with MacTel lose this central vision because the blood vessels in the peripheral regions of the retina start to leak and grow into the macular.
My lab hopes to identify genes candidates for MacTel so that we can design treatments. Our research team is screening animal models with the same retinal vessel abnormalities as human patients to uncover disease genes. Any such discovery would be groundbreaking and thus guaranteed publication in a well-respected academic journal. However, along the way we will have to rule out some genes as not being important, meaning we will generate a lot of ‘negative’ data. This negative data is an important part of the process and a result of the same rigorous scientific techniques that generate positive data. Although it is not considered new, novel or exciting and thus is of little interest to the big names in scientific publishing.
A scientist’s opinion on negative data might depend on where they’re at in their career. A PhD student might see the experiment as ‘not working’; the eager fresh-faced post doc might see it as a waste of their time; but I like to think that the established professor might see it as useful and worthy of publishing.
In my case negative data tells me I must’ve done something wrong. It must be my fault the experiment did not show what we hoped. And to increase my chances of getting a good job after my PhD, I need to get my work published; preferably in a good journal. Thus negative data can cause frustration.
‘The Good, the Bad and the Negative’ aims to challenge the perception of data as ‘negative’. Berit and I wanted to create something as much about science as design. Something we’d both been involved in from its inception, rather than an attempt to simply compress a scientific concept into a pretty design for people to look at. We wanted to comment on the thought processes behind the two disciplines, their similarities, their differences and ultimately what each can bring to the other.
Negative data may not be an exciting answer, but it is nonetheless correct. We should take a step back from our supposed ‘dead ends’ and see where they could actually lead us. It may be different from where we thought we were heading, but who’s to say it wouldn’t be better?
References
Everyone loves games. For eons the act of playing has facilitated our social and mental development. New research has even hinted at gaming as a potential anti-dementia device. Games aren’t just for fun; they’re good for our health.
(Photos by Cléon Daniels)
Most of our brain is cerebral cortex – our thinking mind – a highly folded surface of at least two and half thousand square centimeters. In the outer reaches alone, we lose a staggering 85 000 neurons per day. That’s well over 30 million every year. However, our brains are dynamic and their consistent depletion isn’t necessarily a disaster because they’re quite flexible. This ‘plasticity’ is inherent in the complex network of neurons that ramify the brain, the connections between which can alter. At the junction of two neurons, the more synapses between their contact points, the more evident the neural plasticity. A Label-Based Encephalic ROIs Template (ALBERT) is a Magnetic Resonance Imaging tool that monitors brain development in 3D helping scientists to visualise brain plasticity.
(Photo by Cléon Daniels)
ALBERT in NeuroPlastic Land, unlike Alice in Wonderland, borrows its name from a powerful imaging technology. Though like the fairytale story, it combines imagination with reality to invite the public to play. Elaine Ng (designer) and Ioannis Gousias (scientist) present the ancient Greek game, Triliza, in brightly coloured reflective lenticular plastic that responds to external stimuli. Inspired by the revelations of brain atlasing techniques their installation echoes the plasticity of the human brain.
(Photo by Cléon Daniels)
(Photo by David Nelson)
Ioannis is a Research Fellow at the MRC Clinical Sciences Centre (Imperial College, London) funded by the Action Medical Research and the MRC. Combining brain atlasing with other techniques, he monitors the developing brain ‘network’ and its components throughout life. The aim is to create valid biomarkers of abnormal development or disease.
(Photo by David Nelson)
Elaine is currently taking the MA in Textile Futures at Central Saint Martins College of Art & Design. She combines nature with technology using shape memory material to explore how patterns can respond to changes in environmental conditions such as light intensity or mechanical force. To this end she’s interested in urban textiles and their responses to sun, wind and rain.
Science behind the design
Dr Ioannis Spyridon Gousias reveals the inspiration for ALBERT in NeuroPlastic Land
Three-dimensional atlases and databases of the brain at different ages facilitate the description of neuroanatomy and the monitoring of cerebral growth and development. Paediatric brain atlasing is a currently evolving imaging research field with a wide potential that is beginning to be unleashed. Defining a brain “atlas” as a detailed segmentation of the brain into anatomical regions-of-interest (ROIs), brain segmentation is challenging in young children and neonates due to structural differences compared to adults. Besides, the differences in the characteristics and the properties of the application of Magnetic Resonance Imaging (MRI) in each age group have to be considered.
Conventional MR Imaging is a non-invasive and non-ionising technique and has widely proven its potential for identifying normal and pathologic brain morphology giving objective information about the structure of the neonatal brain during development, since it can be used repeatedly to trace the evolution of a given structure. Brain volumetric studies are able to identify subtle changes that cannot be evaluated by the standard clinical radiological approach.
Manual delineation is the “gold standard” in studies where brain segmentation of MR data sets is required. This can apply both to basic tissue classification and anatomical subdivision. However, it is expert dependent, observer demanding and time consuming, and essentially not scalable. Automated techniques are necessary to overcome these obstacles, especially when large cohorts of datasets are involved, and should approach the accuracy and the details captured with manual delineation. Nevertheless, A Label-Based with Encephalic ROIs Template (ALBERT), in other words a brain atlas, age-specific and manually constructed, can always prove its potential as a prior classifier in age-specific automatic brain segmentation.
The MR image intensities for newborn brains are significantly affected by both low contrast and RF inhomogeneity, which can be difficult to overcome without spatial prior information. Methods that use probabilistic brain atlases or templates for segmentation of healthy adult brain MRI cannot be directly applied to newborn brain MRI since the spatial prior information for rapidly changing myelination property would be very difficult to define. The white matter and gray matter contrast to noise ratio (CNR) for newborn MRI can be as low as half of the one for adult brain MRI. Contrast differences can also be seen in the MR images of neonates and older children. Factors that reduce CNR are the small size of the infant brains and the short scanning period. The small head size requires them to be scanned at higher resolution, which leads to higher noise levels. The infants need to be scanned in a very short time unless they are sedated or constrained. The low CNR causes difficulty in segmenting partial volume regions. One further challenge is that the dividing boundaries between regions that are fully myelinated and non-myelinated are generally ambiguous.
Most studies on neonatal brain segmentation deal with tissue segmentation of cortical GM, subcortical GM, unmyelinated WM, myelinated WM and ventricular space. There are a number of studies that pursue further segmentation of the cortical GM into a small number of regions. However, these segmentations are not based on detailed delineation protocols and do not utilize the full potential of anatomical landmarks as boundary indicators. In order to obtain atlases based on MRIs with characteristics closer to the target group, I have built a database of neonatal ALBERTs which are used as priors for automatic neonatal brain atlasing. These atlases can be used as normal references for assessment of clinically acquired scans and for monitoring developmental changes in the brain. In addition, they can be a useful tool in the monitoring of neuroplasticity, cortical foldings, myelination patterns and developmental growth of different brain regions in longitudinal studies and in group comparisons between normal controls and pathological cases of the same age. They can be combined with functional imaging to determine which anatomical regions are involved in a response in a functional experiment. Finally, they can be used to assess the effectiveness of treatments.
References: