Number crunching – and how you see pink

Data mining and decision-making

Building professionals traditionally relied on years of training and trial and error to build a database of tacit knowledge. But growing complexity, competition and legal liabilities mean that cumulative analogue strategy is just too risky today.

Here are three different approaches harnessing machine learning to stay ahead of the curve.

1. Balancing KPIs

This team used data to model trade-offs between different approaches to automating window shading and artificial lighting (ambient and task) in terms of key performance indicators – energy use and visual performance (light levels and glare). This process enabled a reduction of around 50% in energy use without compromising occupant comfort. Critically, the team found that the benefits were undermined when system was not experienced as trustworthy or reliable, liked to a sense that the locus of control was attributed to a powerful other – Saving energy by maximising daylight and minimising the impact on occupants: An automatic lighting system approach.

2. Tracing causal connections at the population scale

The UK Biobank that contains genetic, lifestyle and health information of over 86,000 adults is an incredible resource for scientists seeking to understand the impact of light on critical aspects of physical and mental health. This study found a direct correlation between risk of major depressive disorder, generalized anxiety disorder, PTSD, psychosis, bipolar disorder, and self-harm behavior in people exposed to low levels of light during the day and bright light at night. The study also notes that, ‘Independent of night light, greater day light exposure was associated with reduced risk for major depressive disorder, PTSD, psychosis, and self-harm behavior’ – Low daytime light and bright night-time light are associated with psychiatric disorders: an objective light study in >85,000 UK Biobank participants.

3. Multivariant analysis

Most scientific research seeks to isolate a single variable and control every other factor to measure an effect. So lighting designers have struggled to know which parameter (brightness, melanopic EDI or duration of exposure) makes the biggest difference to melatonin suppression. This paper mined data from 29 studies to discover just that. They concluded that melanopic EDI dominated the response, followed by light exposure duration, pupil dilation and S‐cone‐opic EDI (spectral power distribution of the light source weighted to the red or short wavelength cones) – Predicting melatonin suppression by light in humans: Unifying photoreceptor-based equivalent daylight illuminances, spectral composition, timing and duration of light exposure.

Enlighten(ed) 

Reflecting on the IALD event last week.

Impeccable organisation, buzzy crowd, stellar line-up of speakers, and a blustery booze cruise and a bop on the Thames… brilliant!

Sadly, only able to stay for one of the two-day event.

But three key takeaways stick in my mind: 

1. AI making waves – digital twins, rear-view mirror predictions, micro-trend-spotting, pulsing illustrations, and new-look selfies. But some tricks from the PropTech world could help us to stay afloat: they are busy integrating adaptive with personal controls across building service platforms and mining sensor data in real-time to drive down the gap between performance on paper and in real life – a stream of arguments we could harness to drive for value over cost.
2. Sustainability – less is more for people, planet and stargazers alike. But as efficiency rises (and when all electricity is ‘carbon neutral’), marginal gains in lumens per watt won’t count for much. Time to widen our focus from luminaire to a systems perspective: the human experience of light in space and time from day into night and through seasons, integrating daylight modelling and harvesting, shading and surface and personal control?
3. Purpose – Every session was glowing with passion, poetry, and attention to detail, sparkling stations and heritage sites. Tentative references to comfort and joy. But no visible sign of the people we are designing for, or reference to the growing scientific evidence that the right light at the right time can help us all feel healthier and happier. Is it time to be more confident about the value of light?

Hot pink under the spotlight 

There is no pink in the rainbow, so how come we see it?

Pink is a brilliant demonstration of how we create colour in our mind’s eye, a century-old theory, currently being challenged by new computational models  with implications for the lighting, especially in schools.

The classic model, known as Colour Opponent Theory, proposes that signals from rods and cones are processed in the visual cortex as opponent pairs  (blue v yellow, red v green and black v white). You’ll see that in action yourself if you stare at these four dots and then look away to a plain neutral surface. If you have ‘normal’ colour vision, you will ‘see’ the opponent colour appear to float before your eyes. That’s because the colour-sensing cells ‘saturate’ and stop sending signals back to the brain, so the opponent colour signal cuts through. (see https://www.allaboutvision.com/resources/human-interest/afterimage/_)  But that’s a long and energy-intensive process and there are some phenomena that Colour Opponency simply can’t explain.

Utility-Based Coding suggests our experience of colour is generated by parallel processing across multiple brain areas based on past experience and context, trading off sharpness and visual signals to give us the critical information we need in the fastest-possible time  Color appearance and the end of Hering’s Opponent-Colors Theory. 

This may explain why lights that deliver a balanced range of wavelengths similar to sunlight (‘full spectrum’) not only support visual comfort, but boost mood, focus and even sleep – Effect of daylight LED on visual comfort, melatonin, mood, waking performance and sleep.

Give peas a chance!

It’s National British Pea Week.

I’ve just sown some in my allotment – no sign of green shoots yet.

It’ll take around 100 days before they’re ready to eat.

So in the meantime, here are three light-related factoids about the humble pea – loads more on the ‘yes pease’ website here.

Eat them raw if you can. Frozen pas are often blanched before freezing because the process breaks down cell walls, releasing natural pigments. But when you boil those peas at home, their nutritional value plummets – Effects of different cooking methods on the chemical and physical properties of carrots and green peas. Cooking ‘sous vide’ is best (here’s how you do that) – Sous Vide Cooking at Home Is Easier Than You Think.

Good for your eyes: Peas contain lutein and zeaxanthin, antioxidants that help protect eyes from harmful blue light and reduce the risk of age-related macular degeneration (AMD) – Effects of lutein supplementation in age-related macular degeneration.

Bio-luminescent Research: Scientists are experimenting with bioluminescent genes in pea plants to study plant health and growth in real-time – A Non-Invasive Approach in the Assessment of Stress Phenomena and Impairment Values in Pea Seeds Caused by Pea Weevil.

And, according to Yes Peas, the world record for eating peas is held by Janet Harris of Sussex who, in 1984, ate 7175 peas one by one in 60 minutes using chopsticks!