Thinking Bugs

Bugs can think. Single genetically identical organisms respond differently to the same stimuli. Large communities rapidly adopt successful response strategies using much faster techniques than classic natural selection. They coordinate group activities using chemical neurotransmitters and electric ion channels. Quorum sensing and ion spiking in bacterial films can be seen as prototype brain activities.

Bugs learn and memorize learned experiences by changing expression of genes in the same way as neurons in our brains do. Single cells design response to earlier unknown threats by changing the topologies of their DNA expression networks, by supercoiling of their DNA.

If we translate physiology and information science vocabulary to the language of psychology and neuroscience we can see learning, habituation, decision making, spiking, long term and short term memory.

It is a philosophical question if we call all the aforementioned processes cognition, computation or just physiology. Those processes in the microbial world are either identical or causally equivalent to the processes in the human brain that we would qualify as intelligence.

Intelligent or not intelligent from the philosophical standpoint thinking bugs represent an excellent model for testing our theories of cognition, intelligence and artificial intelligence, of course. Bugs fire much slower than neurons, on the one hand, and evolve much faster than humans, on the other. They design minimal responses to extremely complex environmental challenges. Biased random walks, causal equivalence groups are only two examples of such suboptimal but working strategies which bacteria applies universally.

Looks like microbiology is preparing a revolution in both neuroscience and artificial intelligence. Yet the tag microbiology has only 260 subscribers at Medium while neuroscience has thousands and artificial intelligence - tens of thousands. A huge untapped opportunity is emerging.

I don't have enough knowledge of the subject yet to write a full fledged article in my own words, but I’ve put together a collection of quotes and links to just some publications which I considered interesting for researchers in both neuroscience and artificial intelligence.

Let me begin with a story about bacterial computer solving an NP- complete problem, that was considered unsolvable for computers so far. Enjoy your journey!

Bacterial Computing

Bacterial Computer Resolves NP-Complete Problem

Solving a Hamiltonian Path Problem with a bacterial computer

“We programmed bacteria with a genetic circuit that enables them to evaluate all possible paths in a directed graph in order to find a Hamiltonian path. We encoded a three node directed graph as DNA segments that were autonomously shuffled randomly inside bacteria by a Hin/hixC recombination system we previously adapted from Salmonella typhimurium for use in Escherichia coli.”

“Bacterial computers can be programmed by constructing genetic circuits to execute an algorithm that is responsive to the environment and whose result can be observed. Each bacterium can examine a solution to a mathematical problem and billions of them can explore billions of possible solutions. Bacterial computers can be automated, made responsive to selection, and reproduce themselves so that more processing capacity is applied to problems over time.”

Seeing the Beautiful Intelligence of Microbes

Somatic Learning, Colony Evolution and Quorum Sensing In Between

Bacterial Cells Learn In Somatic Time And Evolve In Evolutionary Time

Bacterial computing: a form of natural computing and its applications

“Since the different systems of bacterial signaling and the different ways of genetic change are better known and more carefully explored, the whole adaptive possibilities of bacteria may be studied under new angles. For instance, there appear instances of molecular “learning” along the mechanisms of evolution. More in concrete, and looking specifically at the time dimension, the bacterial mechanisms of learning and evolution appear as two different and related mechanisms for adaptation to the environment; in somatic time the former and in evolutionary time the latter. In the present chapter it will be reviewed the possible application of both kinds of mechanisms to prokaryotic molecular computing schemes as well as to the solution of real world problems.”

Bacterial Quorum Sensing — Community-wide Learning

Bacterial Quorum Sensing

“Quorum sensing involves the production, release, and subsequent detection of chemical signal molecules called autoinducers. This process enables populations of bacteria to regulate gene expression, and therefore behavior, on a community-wide scale.”

Brain-like Ion Communication of Bacteria

Bacteria Use Brainlike Bursts of Electricity to Communicate

“Like neurons, bacteria apparently use potassium ions to propagate electrical signals…

Despite the parallels to neural activity, Süel emphasizes that biofilms are not just like brains. Neural signals, which rely on fast-acting sodium channels in addition to the potassium channels, can zip along at more than 100 meters per second — a speed that is critical for enabling animals to engage in sophisticated, rapid-motion behaviors such as hunting. The potassium waves in Bacillus spread at the comparatively tortoise-like rate of a few millimeters per hour. “Basically, we’re observing a primitive form of action potential in these biofilms,” Süel said. “From a mathematical perspective they’re both exactly the same. It’s just that one is much faster.”

Bacterial intelligence: imitation games, time-sharing, and long-range quantum coherence

“Researches on bacterial communication to date suggest that bacteria can communicate with each other using chemical signaling molecules as well as using ion channel mediated electrical signaling…

Besides, the study on bacterial ion channels will give significant insights into the structural basis of neuronal signaling and human brain. Though the quest for the quantum basis of the electrical signaling does not produce the benefit of mankind, but one can’t neglect the underlying theoretical aspect of the quantum viewpoint.”

Other Bacterial Learning Strategies

Single-cell Organisms Learn By Habituation

“Her group not only taught slime molds to ignore noxious substances that they would normally avoid, but demonstrated that the organisms could remember this behavior after a year of physiologically disruptive enforced sleep.”

Slime Molds Remember — but Do They Learn?

Habituation in non-neural organisms: evidence from slime moulds

Bacteria Habituate To Toxins Somatically

Tolerance response of multidrug-resistant Salmonella enterica strains to habituation to Origanum vulgare L. essential oil

Acid habituation of Escherichia coli and the potential role of cyclopropane fatty acids in low pH tolerance

Bacteria Learned To Feed On Antibiotics

Shared strategies for β-lactam catabolism in the soil microbiome

How Bacteria Eat Penicillin

Bacteria Learned To Use Old Method Against New Threats

Bifunctional Immunity Proteins Protect Bacteria against FtsZ-Targeting ADP-Ribosylating Toxins

Signalling Networks

Signal Transduction: Networks and Integrated Circuits in Bacterial Cognition

“The ultimate level of integration in bacterial regulatory networks occurs on DNA. Activation of one transcription factor generally leads to changes in the expression of numerous genes including other transcription factors. These changes feed back to modulate the signal transduction pathways that generated them in the first place.”

DNA Methylation and Memory Formation in Bugs…

Importance of Multiple Methylation Sites in Escherichia coli Chemotaxis

“Bacteria navigate within inhomogeneous environments by temporally comparing concentrations of chemoeffectors over the course of a few seconds and biasing their rate of reorientations accordingly, thereby drifting towards more favorable conditions. This navigation requires a short-term memory achieved through the sequential methylations and demethylations of several specific glutamate residues on the chemotaxis receptors, which progressively adjusts the receptors’ activity to track the levels of stimulation encountered by the cell with a delay.”

Memory in Microbes: Quantifying History-Dependent Behavior in a Bacterium

“In cellular systems, environmental memory has been noted to be inherent in everything from the selective history of mutation, epigenetic inheritance via chromatin modification in neurons and DNA methylation in chemotaxing bacteria, genetic and epigenetic phase variation mechanisms controlling surface features of pathogenic bacteria, cellular proliferation and survival in the immune system, and in switch-like feedback systems in regulatory networks spanning signal transduction, metabolism and gene expression.”

… and Brains

DNA methylation and Memory Formation

“Memory formation and storage require long-lasting changes within memory-related neuronal circuits. Recent evidence indicates that DNA methylation may serve as a contributing mechanism in memory formation and storage.”

Dynamic DNA methylation in the brain: a new epigenetic mark for experience-dependent plasticity

“Epigenetic modifications of histone proteins and DNA seem to be a leading molecular mechanism to modulate the transcriptional changes underlying the fine-tuning of synaptic connections and circuitry rewiring during activity-dependent plasticity.”

Variable DNA Topology Shapes Gene Expression Response to Unpredictable Environmental Changes (Constrained by the topology of DNA networks)

DNA supercoiling is a fundamental regulatory principle in the control of bacterial gene expression

“The topological state of DNA also influences its affinity for some DNA binding proteins, especially in DNA sequences that have a high A + T base content. The underwinding of DNA by the ATP-dependent topoisomerase DNA gyrase creates a continuum between metabolic flux, DNA topology and gene expression that underpins the global response of the genome to changes in the intracellular and external environments. These connections describe a fundamental and generalised mechanism affecting global gene expression that underlies the specific control of transcription operating through conventional transcription factors. This mechanism also provides a basal level of control for genes acquired by horizontal DNA transfer, assisting microbial evolution, including the evolution of pathogenic bacteria.”

“Variable DNA topology serves as the basis of a global regulatory system that allows the gene expression profile of the bacterium to vary in response to those environmental influences that cause DNA topology to change (Dorman 1991, 2006). Shifts in the ratio of ATP to ADP can arise from a very wide range of circumstances… Exploiting the topological state of the DNA in the chromosome as both a sensor and a gene-to-gene telegraph of physiological status allows the individual cell to adjust its transcription profile to ensure an optimal response to changing circumstances. Variations in the quality and quantity of the response between individual cells creates variety across a population of genetically identical bacterial cells that allows differences in relative competitive fitness to emerge that may prove useful in the survival of the population.”

Bacterial genome architecture shapes global transcriptional regulation by DNA supercoiling

“DNA supercoiling acts as a global transcriptional regulator in bacteria, that plays an important role in adapting their expression programme to environmental
changes, but for which no quantitative or even qualitative regulatory model is available. Here, we focus on spatial supercoiling heterogeneities caused by the
transcription process itself, which strongly contribute to this regulation mode. We propose a new mechanistic modeling of the transcription-supercoiling dynamical coupling along a genome.”

“Simulations show that these global expression responses to changes in DNA supercoiling result from fundamental mechanical constraints imposed by transcription, independently from more specific regulation of each promoter. These constraints underpin a significant and predictable contribution to the complex rules by which bacteria use DNA supercoiling as a global but fine-tuned transcriptional regulator.”

A Microarray-Based Antibiotic Screen Identifies a Regulatory Role for Supercoiling in the Osmotic Stress Response of Escherichia coli

“Because long-term survival is contingent on adaptation to a new environment, transcriptional activity of the genes found in cluster 0 may be hardwired for optimal expression at high levels of negative supercoiling.”

Gene Regulatory Networks Adjust Physiology To Flexibly React to Environmental Changes (constrained by network topology)

Adaptation of cells to new environments

“Environmental adaptation of biological systems can be considered from three evolutionary perspectives: (i) acclimation of existing cellular machinery to operate optimally in a new environmental niche; (ii) acquisition of entirely new capabilities through horizontal gene transfer or neofunctionalization of gene duplications and (iii) reorganization of network dynamics to appropriately adjust existing physiological processes to match dynamic environmental changes.”

“Systems-level coordination of cellular functions is accomplished by gene regulatory networks (GRNs)”

Predictive behavior within microbial genetic networks

“While rewiring of GRNs is an efficient mechanism to acquire new features or functions, it is important to remember that preexisting network topology constrains the space of viable and visible phenotypic outcomes that can result from alterations to its structure (59), especially over short time scales.”

“Randomly evolving biochemical networks of these organisms form internal representations of their dynamic environments that enable predictive behavior. We provide experimental evidence for this capacity by revealing strong correlations in genome-wide transcriptional responses of E. coli to transitions in oxygen and temperature.”

“We show that in silico biochemical networks, evolving randomly under precisely defined complex habitats, capture the dynamical, multi-dimensional structure of diverse environments by forming internal models that allow prediction of environmental change.”

Causal Sets (Classes of Functional Equivalence) Provide Alternative Topologies of Gene Regulatory Networks

Downward causation by information control in micro-organisms

“The concepts of functional equivalence classes and information control in living systems are useful to characterize downward (or top-down) causation by feedback information control in synthetic biology. Herein, we re-analyse published experiments of microbiology and synthetic biology that demonstrate the existence of several classes of functional equivalence in microbial organisms. Classes of functional equivalence from the bacterial operating system, which processes and controls the information encoded in the genome, can readily be interpreted as strong evidence, if not demonstration, of top-down causation (TDC) by information control.”

“A class of functional equivalence is defined by a functional outcome (goal) that is operated by lower level components that can be different as long as they produce the same outcome. By exemplifying the conservation of functions rather than the conservation of modes of operations, the existence of classes of equivalence strongly suggests that it is the biological system as a whole that defines the boundaries and constraints within which a particular class of functional equivalence is established by natural selection. In other words, it is a functional need developed by the whole biological system that defines the constraints within which a particular class of equivalence is established.”

Optimized Bacteria are Environmental Prediction Engines

“We find that the instantaneous predictive information — that shared between the organism’s present phenotype and future environment states — captures (and not just upper bounds) the benefit of epigenetic memory. When combined with resource constraints, this predicts that optimal isogenic bacteria populations store epigenetic memories that are causal states of bacteria observations of the environment. We conclude with suggestions for testing and extending these results.”

“To show this, we first show that expected log growth rate is maximized when the epigenetic memories store the entire observed environmental past. Then, we show that this maximum is also achieved when epigenetic memories are minimal sufficient statistics of prediction of the future environment with respect to past observations. Finally, the aforementioned resource constraints imply that optimal realizable epigenetic memories are causal states.”

Biased Random Walks Are the Basis of (not only) Bacterial Behavior

Natural Agency: The Case of Bacterial Cognition

“To regulate the tumbling frequency, bacteria exploit a pervasive phenomenon in nature: random walks. These are processes in which moving objects, such as molecules, have an equal probability of moving in any direction away from their starting point. The particles composing a gas can serve as an example; they bounce around in a container and change their direction as they collide. What is distinctive about chemotaxis — and metabolism more generally — is that cells can bias these random walks in a direction that is on average adaptive. Unlike the movement of inanimate particles, the movement of bacteria is adaptively rather than randomly self-directed. Moving toward the source of attraction is thus not something that just passively happens to a bacterium as a blind mechanical consequence of genetic and environmental events. Changing the rate of tumble appropriately is rather something that the bacterium itself actively does. Bacteria are thus agents guided by the relevance that the chemical content of their environment has for fulfilling their life cycle.”

The Weierstrassian movement patterns of snails

“We also show in an easily accessible way how chaos can produce a wide variety of Weierstrassian Lévy walk movement patterns. Our findings support the Lévy flight foraging hypothesis that posits that because Lévy walks can optimize search efficiencies, natural selection should have led to adaptations for Lévy walks.”

“Our analysis thereby shows that Weierstrassian Lévy walks can arise for ‘free’ from generic properties of chaos and does not require sophisticated internal rules governing the switching between a variety of random walks each with its own characteristic step length.”

“it is striking that the composite Brownian walks were finely tuned to theoretically optimal Lévy walks, suggesting selection pressure for Lévy walk characteristics. Our findings suggest that this congruence is not unexpected given the presence of chaos.”

Bacterial Future Of AI

Macromolecular networks and intelligence in microorganisms

“Here, we explore how macromolecular networks in microbes confer intelligent characteristics, such as memory, anticipation, adaptation and reflection and we review current understanding of how network organization reflects the type of intelligence required for the environments in which they were selected…

Large macromolecules, such as proteins and polynucleotides, may store information as, for example, Gibbs free energy in metastable states, where interactions between their structural components can differ depending on the way they were folded some time ago…

Given the examples of the previous section, it is likely that, at least for some specific tasks, microbial “intelligence” can be compared to human intelligence, and microbial networks could be considered formally as “intelligent…”

Intelligence is a strongly emergent property in both microorganisms and animals, including humans. Still, there is a difference in the way these intelligences are manifested. Thus, humans study microorganisms and debate about microbial intelligence, and bacteria, while supremely adapted and aware of their environments, are probably not even aware of us and our endeavors.”

Is Smaller Better? A Proposal to Use Bacteria For Neuroscientific Modeling

“The complexity of neuronal function has hindered the rate of progress in biologically inspired modeling. For example, despite the general lack of understanding of neuronal network function, artificial neural networks (ANNs) still use as their basic computational unit a binary input/output node that is meant to be an abstraction of a neuron (Lippmann, 1987). However, this is an oversimplified view of actual neuronal function, one that prevents the network from being able to fully harness the abilities of biological neuronal networks in order to aid computation (Staelin and Staelin, 2011).

In this review, we begin with an overview of similarities between mammals and lower-order organisms, and proceed to a discussion of the ways in which bacterial communities mirror neuronal circuits and networks. By focusing on these similarities, we hope to motivate improvements in models of higher-level organisms such as mammals and to influence biologically inspired computational efforts such as ANNs.”