There's a whole lot wrong with this paper, but I think that the biggest is that it is attacking a giant fantasy—the idea that biology above the molecular level has been abandoned. As well as the fact that the subfields of biology that aren't molecular biology and biochemistry still exist as active fields of study and research, a lot of what people working in molecular biology and genetics are doing is looking at the relationship between low-level processes and higher-level structure and activity.
This. The authors point is analogous to "we shouldn't do the census because everything is a result of human activity, and humans are really smart". Well... people study psychology and others study economics. As computational and modeling power goes up we have the newfound ability to blend the micro and macro further. But historically this wasn't the case.
Kind-of a bummer article for what I think is a very good point (that I'm not actually sure he is trying to make). Current models based on biological principals are starting at a level of complexity which is emergent from a more foundational level which is ignored in our currently best performing models.
There is a strong argument for understanding 'mind' or 'intelligence' at a cellular level. Neural networks were originally inspired by how we thought brains work; a highly connected network. Within this framework, it was kind-of ignored that the fundamental units of this network themselves have 'mind' (intelligence? sense of self?) and at least the ability to change their behavior (outputs) based on what they sense from their environment (inputs). Currently, we train NN based on training data, but once trained, the network is 'fixed', in that we've come up with some optimal set of weights and biases that don't change. Biological networks don't work like this. Individual units within biological networks can individually adjust their behavior, based on external inputs both from other cells in the network and external to whole network (environmental) inputs. I would argue that there has been a strong bias in cell biology (especially human cell biology) against attributing individual cells with 'mind' or a sense of self; however, this obviously can not be the case, since most of the diversity of life is unicellular.
I think there would be a strong benefit to considering the ideas of 'cells' and compartmentalization in regards to how we design and engineer intelligent systems: we know that fundamentally, this is at least part of how cells work.
> Current models based on biological principals are starting at a level of complexity which is emergent from a more foundational level which is ignored in our currently best performing models.
Not really. It's too complex, we understand too little, and there's very little reason for it, since intelligence is not a property which is encoded in a neuron, but an emergent property, emerging from the interactions between them and the development of the network.
It's a bad idea for the same reason that if you want to simulate the aerodynamic properties of an airplane, you don't start with string theory.
> Individual units within biological networks can individually adjust their behavior, based on external inputs both from other cells in the network and external to whole network (environmental) inputs.
You don't need to start from the cellular level to achieve this though.
Wow. Cells are machines made of molecules. Saying we shouldn't study molecular biology is like saying that it's a waste of time for electrical engineers to understand transistors.
I think he's misunderstood reductionism too. Reductionism isn't about saying "only the lowest level detail matters", it's saying "the lowest level detail explains the higher levels".
True. But only if the "levels" are identified correctly.
Using the language of s/w: in your design, you have to define proper layers - not too many, not too few. As opposed to trying to express everything directly in terms of assembly codes, without formalizing the intermediate concepts.
The author's point (IMO) is that if we want to learn anything about our intelligence, we need to explore an obvious path: study the intelligence of cells first.
This doesn't eliminate the importance of other levels. There can be, in turn, something inside the cell that is crucial to its "intelligence", who knows? But first thing first :)
Its pretty clear that the author of the paper comes from either a non-scientific or a barely scientific background on writing style alone, but I think in broad strokes, the point absolutely stands, and its a point that keeps me up at night.
Intelligence (or 'mind') is an emergent property of cells (or maybe even of compartmentalization). While its clearly been useful to study intelligence a level above cells (networks), in biological systems, multicellularity and thus networks are emergent properties of individual cells. We currently ignore the function of individual cellular intelligence in the models we develop based on biological frameworks.
Anyone interested in a more scientifically sound treatise of cellular biology and from a computational perspective should instead read "Wetware: A computer in every living cell" by Dennis Bray. It's very approachable and, despite its unfortunate title, has nothing to do with the absurd anthropomorphising suggested in the posted Brian Ford paper.
Much of the money spent on increased resolution in electron microscopy is wasted? What a crackpot indeed. This was written before the resolution revolution, but still...
I can't recall now where I read this, but it's a very interesting idea: Your life is the life of your cells.
Everything we can do is something cells can do writ large.
Senses: we can see because our cells can see. Likewise for taste, vibration, heat/cold, all twenty-seven of them (I think we're up to twenty-seven human senses.)
Digestion? Cells. Movement? Skeletons? These are all things our cells do.
There is nothing we do that is not done by our cells.
If cells operate in a similar fashion, where adaptation to externalities encodes itself into a durable surface of arbitrary pockets, then we have a recipe for inert systems exhibiting variable degrees of behavior, which is capable of evolving over time.
This doesn't solve for sentience, but it does solve for expressive adaptation to fit into complex, hostile environments.
Initial thoughts on reading the abstract: what a crackpot.
There's a good reason to sub-cellular structures and interactions: we need to have a reductionist's understanding in order to better understand the whole. We never would've had any real insight into the cell, or much of our modern day medicine, without all those researchers putting in hard, arduous hours into studying the molecular mechanics of single proteins.
Ford would apparently have us see the cell as a antropomorphised black box, attributing all behaviour to 'cellular intelligence'.
If you read the text, you'll find that he calls for biologists to examine whole cells and systems of cells as well as the pieces thereof. That's actually a legitimate request for balance in the field.
While I am skeptical of attributing personality changes due to cell memory from organ transplanting, is it possible for a cell or a cell collective to have information (memory) embedded as a dynamical system? Theoretically, it may be possible to embed memory into a dynamical system [1].
He suggests that cells are intelligent, and have a memory, and that this is the reason why a middle-aged man started liking classical music after having a donor organ from a violinist.
Cells are intelligent. And have a memory. I'm a microbiologist, studying bacteria (single cells), and let me tell you - these things are true.
The organ transplantation thing seems far fetched and probably BS, but the idea that someone's mood could change after an organ transplant is not that far fetched. My mom studies transplantation and I know a bit of immunology as well. Organ transplantation stimulates an immune response and in turn release of various cytokines into the blood, which could travel to the brain and affect mood. Inflammation in the brain is linked to many mood disorders.
The worst thing about the article is it's way too long, I think the author must like to hear himself talk. I'm also not sure of why he dismisses quorum sensing. Surely the "whole" does not stop at the single cell level. There's the whole community, etc..
He didn't seem talking about genomics or epigenomics, though.
> Cells are intelligent.
Define intelligent. I don't see a mechanistic response unit, no matter how sophisticated, as intelligent.
> Inflammation in the brain is linked to many mood disorders.
That is, however, extremely different from starting to like classical music because your donor was a violinist. He wasn't talking about mood disorders, but specific preferences.
Upon re-reading the organ transplant section, you're right that he says some crazy stuff. However, what he says should be experimentally testable, which means it's a valid hypothesis. I don't think the evidence he provided is anywhere near sufficient to support his claim though.
> I don't see a mechanistic response unit, no matter how sophisticated, as intelligent.
Can you provide an example of an "intelligent" system that is definitely not a "mechanistic response unit"?
> He didn't seem talking about genomics or epigenomics, though
Proteins can hold memory too. Not as stable as DNA except maybe for amyloid fibers like those in prions. But it's not out of the question.
> Can you provide an example of an "intelligent" system that is definitely not a "mechanistic response unit"?
Sure, intelligent systems do by definition have a mechanistically responsive element at its base; they're made of matter, after all. However, that doesn't mean that every system which is able to respond to changes in its environment is ipso facto intelligent.
The neurons in the brain are not what makes a brain intelligent. It's the emergent meta-structures formed by neurons which allow us to think the way we think.
Firstly is the idea that we somehow diminish a biological system by viewing it in a reductionist manner. I believe this to be patently false. This deeper understanding, to me, at least, only serves to enhance my awe of biological systems in general. Furthermore, as dragonwriter wrote, the author strongly seems to imply that all research on the level of organisms and above has been abandoned completely. I have quite a few ex-colleagues who'd beg to differ.
Secondly, I'm quite bothered by the antromorphisation and attribution of intelligence which happens in the article. Terms like 'undeniable ingenuity' get thrown about with some regularity. I'm not a fan of this, firstly since it's quite clearly a rethorical device to convince readers rather than a rational argument. However, I think the entire premise is based on redefining intelligence to something which I feel intelligence is not. To quote: 'in defining intelligence, adaptation of and to the environment, reaction to unforeseen circumstances and communitaction others is frequently mentioned. It is the essence of such intelligence that we observe in single cells.'
I think this is a highly flawed definition, although I do have to agree with the author that creating a good definition of it is highly contensious. However, I do believe that the ability to analyse ideas and to evaluate their impact on future events is a core component on this.
I would say that the behaviour defined as 'intelligent' or 'skillful' in this article, such as the healing of injuries or light sensitivity of stomata are misnomers. To describe such behaviour as intelligent is attempting to mystify what is (to me at least) a process of mechanistic adaptation.
Then, to my third point, the author goes from vague mystification straight into pseudoscientific theory such as cellular memory, the section of which is fairly underrefferenced for something described as a 'well defined phenomenon'. After that, he does the bold claim that 'the brain's capacity fundamentally redies in the intraneuronal data processing rather than mere interneuronal activity', citing for this that neurons change when exited regularly. In other words: brains do not cause intelligence because of neuronal interactions, but because neurons themselves already are intelligent. He then goes on about how he found the modulated sound of a neuron's spike activity 'hypnotic' and having voicelike quantities. He then makes the vexing statement 'The clear impression is that we are listening to the discrete signals with which one neuron in some sense addresses another'.
This statement should not be vexing. However, the intention behind it makes it so. The entire purpose of neuronal spike activity is to propagate information to the next neuron. So it is very much the descrete signal with which a neuron addresses others. However, the author clearly intended this sentence to convey the idea that the cell itself is holding some intelligent conversation with a neighbour. This man is a published biologist; how in the hell did he manage to publish this? Did he still have tenure at Cambridge after submitting this? Because I would hope that he'd get some pointed questions and strange looks on his next performance review.
Lastly, there is the language. As I mentioned earlier in this rambling rant, Ford seems more interested in using cheap rethorical tricks such as appeal to authority rather than rational arguments. I find the tone and content to be entirely inappropriate for a article published in something that pretends to be a scientific journal (although the title, description and impact factor of 0.4 all seem to dispute this).
All in all, this article managed to push all of the many buttons I have.
This was written in 2009. Perhaps one can be kinder to the author in his pleading that looking at cells as individual entities was not getting adequate research focus. Also, things like changes after organ donation do not yet have any scientific explanation.
> This was written in 2009. Perhaps one can be kinder to the author in his pleading that looking at cells as individual entities was not getting adequate research focus.
It wasn't that underresearched in 2009 either, however. Microbiology has always been an active field.
> Also, things like changes after organ donation do not yet have any scientific explanation.
> This was written in 2009. Perhaps one can be kinder to the author in his pleading that looking at cells as individual entities was not getting adequate research focus.
What does the first sentence have to do with the second? What's
changed since 2009 that would be relevant?
