Is Our Framework for Lighting Sufficient?

Part 1: Concepts

In this three-part post, I share some practical examples in lighting where open questions remain.

In each of these problems, some suggest that our standard framework for thinking about light is insufficient and a paradigm change is needed to find better explanations. Others claim that these discrepancies do not exist and we can explain all observable phenomena well enough.

Here, I argue that open questions do exist and we should face them and actively look for answers. Both within and outside of the current paradigm — since we don't yet know where the answers lie.

At some point, all frameworks/models turn out to be insufficient — demonstrating this faster means faster progress.

The Model

In lighting, we use a fairly simple framework to model, describe, and measure light. 

According to this, light has properties like luminance, luminous flux, luminous intensity, illuminance, spectral power distribution — and a few more, depending on the field —, and it follows some geometrical principles.

Okay, strictly speaking, that's not really true... These properties only describe illumination and light is this mysterious thing that we cannot grasp directly. Since something has to cause illumination, we have various theories and models for light, even if none of them seem complete (infamously resulting, e.g., in the wave-particle duality).

But those who stay on the practical level of lighting do not have to care about that... at least for now. We can simplify things by calling illumination "light" and by describing it with properties like in the list above.

These should cover everything we need to find correlates with effects on humans and then to uncover the causal chains of various interactions. Right?

How Do We Know What Property We Might Be Missing?

Empirical testing in lighting starts with something like this: we have two light sources (A and B) on one side, some observers on the other (research subjects of any kind — let's stick with humans), and we compare the effects of A and B on the observer along some psychological or physiological measure in a controlled situation. 

Here, light properties are not primary. What we can know directly are A and B, the response in observers, and of course the situation that we set up.

Light properties are already secondary; parts of the model through which we try to find the mechanisms and how the connection between stimulus and reaction happened. In most experiments, we are already past this stage and go for the details, making the properties (which are dependent on our model!) our independent variable. This is fine and necessary for digging deep into the mechanistic details within our model — but those who are only exposed to these kinds of experiments are more likely to miss the forest for the trees.

In other words: relative to measuring the effects of A vs. B (lamp, daylighting condition, etc.), this introduces a new set of assumptions, which can be incorrect, incomplete, or not applicable to the given test.

It seems like I'm just splitting hairs. But failing to see this distinction results in not looking at potentially interesting questions and severely limits the chance of a new discovery.

Basically, if we think that our model is correct, then all we do is keep filling the gaps in the model — instead of looking for a better one. What's more, if a better model shows up, we won't even recognize it, as it is not what we were looking for.

Some Recent Examples in Lighting

1 The "Blue Light" Topic and ipRGCs

In the early 2000s, the discovery of intrinsically photosensitive retinal ganglion cells (ipRGCs) opened up a whole new field of dealing with the circadian effects of short wavelength light.

These effects were not only there before (since the dawn of times), they were also clearly observable and demonstrated. But the consensus model for thinking about light had no place for them, so they were ignored until a clear mechanism was presented.

Before this transition, vision-based metrics (like luminosity functions) were not sufficient to explain stimulus-response relationships. After a paradigm change, new measures (like mEDI) and a different framework for thinking about light were adopted.

2 Near-Infrared and Health

For decades, many hypothesized that near-infrared light might have health effects that we are missing if we exclude it from our (indoor) lives.

Despite thousands of papers showing interactions on the level of basic biology and in clinical practice, up until a few years ago, you were called woo-woo if you even mentioned the topic in lighting. This was still the case in 2019, when I published a short comment on the question in Lighting Research and Technology (to my knowledge, a first, in terms of acceptance in a mainstream academic lighting journal). Professors reached out privately to say that they agreed — but wouldn't talk about it in public. At a lighting conference, a highly regarded expert in daylighting winked knowingly, hinting at the fact that what matters is not about visible light — and that we shouldn't talk about it here.

Somehow, the trend has changed and the infrared topic is becoming acceptable. This was not due to any new discovery — perhaps just reaching a tipping point or maybe the influence of business potential. 

In this example, I don't think that we have a solid (mechanistic) framework to explain the effects. Maybe the usual properties of light and the interactions described through them could explain these phenomena, but so far, they haven't really. This has certainly been a force to hold back the adoption of the idea in the academic mainstream.

3 Polarization and Eye Strain

When it comes to displays, we care about luminance, contrast ratio, spectrum (usually just color depth and gamut), temporal properties, reflectance, viewing angles, resolution, maybe homogeneity.

LCDs need to be polarized, but it's not part of the list of properties that everyone has been looking at when it comes to ergonomics. (Arguably, spectrum has also been excluded from that list...)

It turns out, polarization may have to do with comfort, perhaps even health. Still early days, mechanisms may not be clear, but why not look there? We've been using polarized displays for decades; this idea shouldn't be anything new! Yet, here we go again — out of sight, out of mind.

Outlook: Open Questions

In the next two posts, I'll break down some potential inconsistencies between commonly observed effects of light and our standard framework for lighting.

These may end up being similar to the examples above. There are two options:

1. They can be explained with some known property, but we are not looking closely enough.

2. We don't even have the right concepts in our standard framework to know how these effects might take place.

Either way, both examples show current conundrums — yet, the consensus says: "nothing to see here." Meanwhile, lots of evidence suggests that these problems are exactly what we should be studying, if we want to learn something new.

Part 2 will be about a spectral problem; Part 3 about a spatial problem. But the main point is epistemological: we take our framework for thinking about light for granted and as if it was complete. But how do we know where its limits and blind spots are? We should at least try to look.