Fortunately, there is more than one kind of “down.” In science and engineering, the usual down involves a search for a system’s smallest and most fundamental components. But the dive for fundamental components, such as atoms and molecules, can descend past things that are fundamental in their own right, namely, the emergent properties of complex systems.
Emergent properties cannot be understood—in practice or even in principle—by identifying and characterizing individual parts. Instead, rules of interaction must be uncovered. But how? According to Nobel Prize–winning physicist Robert Laughlin, what’s needed is a bottom-down approach, which sounds paradoxical but really just refers to a different kind of down, one that gets to the bottom of what we might call interaction spaces.
In A Different Universe, Laughlin argued that physics needn’t focus too intently on fundamental particles. Instead, it can recognize the importance (or even primacy, in some contexts) of the interactions between simple components, and thus account for otherwise unexplainable emergent phenomena such as phase transitions and superconductivity. Something similar could be said of biology. It needn’t focus too much on genomic details. Instead, it can devote a large share of its attention to the interactions that pertain at higher levels of organization. For example, it can uncover the rules that explain how cells self-organize into tissues.
Sometimes, it is necessary to recognize when events at the level of cells or cell collectives are of central importance, despite the excitement over the possibilities offered by genomics. This view goes back as far as 2004, when microbiologist and biophysicist Carl R. Woese, most famous for defining the Archaea, offered his views on what “a new biology for a new century” would entail. Even though the Human Genome Project had been completed just a year earlier, Woese wrote that, for biology, “molecular paradigm [had] run its course,” and would be most suitable for engineering purposes. He added that if biology were to become society’s teacher and not be reduced to doing society’s bidding, it would need a more holistic paradigm, one that would consider factors such as “evolution, emergence, and biology’s innate complexity.”
Were Woese still with us, he might be surprised that an overly reductive molecular paradigm is becoming less and less satisfactory not just in biology proper, but in disciplines that blur the distinction between biology and engineering. For example, synthetic biology, a discipline devoted to improving the engineerability of living systems, mainly individual cells, is broadening its scope to encompass synthetic morphology, which seeks to control the forms and functions that can be assumed by cellular collectives.
As synthetic biology expands its remit from individual cells to cellular collectives, including tissues and organs, it is expanding its collection of tools. At present, most tools consist of genes and gene networks that are meant to give cells new abilities, such as the ability to consume unconventional feedstocks, secrete nonnatural compounds, and respond to physiological or environmental conditions (perhaps by serving as recording or signaling devices). But additional tools are being developed to influence how cells communicate with one another, typically via the modification of chemical, mechanical, and electrical signals.
Prospects for such tools were discussed in an early paper about synthetic morphology. The paper, written by the University of Edinburgh’s Jamie A. Davies, was published in 2008 and called for the creation of a library of sensor, regulatory, and effector modules. Connected functionally within cells, these modules could, Davies suggested, program cells to organize themselves into “specific, novel arrangements, structures, and tissues.”
Davies noted that the beginnings of such a library already existed. For example, he identified master regulators that could activate morphogenetic modules for 10 basic cellular events: apoptosis, cell proliferation, cell fusion, cell locomotion, chemotaxis, haptotaxis, cell–cell adhesion/condensation, cell sorting, epithelial-mesenchymal transition, and epithelial folding.
Davies and colleagues soon extended this work. In 2014, they reported that they had constructed inducible effector modules for the control of cell adhesion, elective cell death, cell fusion, cell locomotion, and proliferation. They also described how these modules were used in a human cell line to induce morphological behaviors on command.
“[The constructed modules] are produced in a standardized format, easy to connect to existing logic and sensory modules,” the researchers detailed. “We also report the testing of these synthetic, inducible modules in mammalian cells.” They added that they hoped to “extend engineered control of morphogenetic cell behavior to more complex 3D structures that can inform embryologists and may, in the future, be used in surgery and regenerative medicine, making synthetic morphology a powerful tool for developmental biology and tissue engineering.”
In the course of his research, Davies began referring to the “rules” of signal-driven behaviors, and to the cells and cell collectives that follow the rules as “agents.” For example, in a public lecture at the University of Edinburgh in 2010, Davies said that form and function emerged in kidneys through the action of “individual agents following simple rules in a dumb way to do something remarkable.”
Developing this idea further, Davies, along with Tufts University’s Michael Levin, described how synthetic morphology could be accomplished with agential materials. In a 2023 paper, they suggested that design strategies could use agential materials in addition to traditional gene- and protein-based engineering approaches. Essentially, the authors envisioned a collaborative approach, one in which designers cooperated with agential materials, which the authors recognized as having their own goals, agendas, and powers of problem-solving.
“Such an agential bioengineering approach could transform developmental biology, regenerative medicine, and robotics, building on frameworks that include active, computational, and agential matter,” Davies and Levin wrote. “We highlight experimental embryology studies … that go beyond the familiar, default outcomes of embryogenesis, revealing the plasticity, interoperability, and problem-solving capacities of life.”
Ideas like those expressed by Davies and Levin are being pursued by scientists at MIT’s Multi-Cellular Engineered Living Systems (M-CELS), which was launched in 2020. Whereas Davies and Levin refer to agential materials, M-CELS refer to “purpose-driven living systems with multiple interacting living components.”
M-CELS’s three founders—Roger D. Kamm, Linda G. Griffith, and Ron Weiss—all contributed to a 2018 article on the promise of multicellular engineered living systems. According to this article, M-CELS can be fabricated in at least two different ways: “top-down” assembly (such as building two- or three-dimensional forms and seeding or plating them with differentiated stem cells or primary cells) or “emergent engineering” (pluripotent cells subjected to guidance cues to drive differentiation and self-organization). The article also described M-CELS applications: organ-on-chip or tissue-chip systems (for disease modeling and drug discovery), implantable “hyper-organs” (for sensing biological signals and synthesizing and secreting a biologic product in response), and biological actuators or bio-robots (for drug delivery and microsurgery).
Progress toward such applications was evident at “M-CELS: Advances in Basic Research and Translational Opportunities,” an M-CELS-organized symposium that was held March 26–28, 2025, at Hilton Head, NC. At this symposium, most presentations described advances relevant to specific application areas, such as the following:
Vascularized organoids: The Wyss Institute’s Jennifer A. Lewis described how human induced pluripotent stem cells can be transformed into vascularized kidney organs via scalable differentiation protocols. She also reviewed her group’s recent work on embedding vasculature into human kidney tissue models via coaxial sacrificial writing in functional tissue (co-SWIFT), a bioprinting method capable of generating hierarchically branching, multilayered vascular networks within both granular hydrogel and densely cellular matrices. Earlier, in an announcement that accompanied the journal article describing this work, Lewis stated, “We were able to successfully 3D-print a model of the vasculature of the left coronary artery based on data from a real patient, which demonstrates the potential utility of co-SWIFT for creating patient-specific, vascularized human organs.”
Vascularized models were also discussed by UC Irvine’s Christopher C.W. Hughes. His laboratory has developed vascularized micro-organs emulating the liver, pancreas, lung, and brain (which includes a blood-brain barrier) as well as vascularized micro-tumors. The technology is currently licensed to Aracari Biosciences, which Hughes co-founded and now serves as CSO.
Innervated tissue models: Binghamton University’s Tracy Hookway discussed the engineering of innervated multicellular cardiovascular models, using pluripotent stem cells that have been differentiated into sympathetic and parasympathetic neurons. In Hookway’s work, neuronal cell populations are mixed with cardiomyocytes at different developmental stages to investigate their influence over beat rate, contraction force, and calcium-handling kinetics.
Immune organoids: Georgia Tech’s Ankur Singh discussed the development of synthetic, human ex vivo immune organoids to replicate the structure and function of immune tissues. In recent work, Singh’s group developed synthetic hydrogels mimicking the lymphoid tissue microenvironment, enabling human germinal centers from tonsils and peripheral blood mononuclear cell–derived B cells. In an announcement pertaining to this work, Singh stated, “Our synthetic hydrogels [allow] us to model antibody production from scratch, more precisely, and for a longer duration. This is a gamechanger for understanding and treating immune vulnerabilities in patients with lymphoma who have undergone cancer treatment—and hopefully other disorders too.”
Embryoid models: The University of Pittsburgh’s Mo Ebrahimkhani , a researcher attuned to the cellular cross-talk in advanced organoid models, described the heX-embryoid, an embryoid model that is derived from adult cells and shows self-organizing peri/post-implantation cellular programs, including the formation of the amniotic cavity and body axis generation. To date, the early post-implantation stages of human development have been difficult to study due to technical and ethical challenges. In press materials that accompanied heX-embryoid’s introduction, Ebrahimkhani asserted that the new technology would “unlock this ‘black box’ of human development, which could help solve the mystery of why about 60% of pregnancies fail in the first two weeks.”
Patient-specific multiorgan models: Columbia University’s Gordana Vunjak-Novakovic discussed her group’s development of organs-on-chip models of human pathophysiology. For example, she highlighted a multi-organ chip that incorporates different tissue niches that are separated by endothelial barriers, yet linked by a vascular flow containing circulating cells.
In a recent article, Vunjak-Novakovic and colleagues asserted that their models support communication between tissues while preserving their individual phenotypes. In a statement that accompanied the paper, she noted, “We can model a patient’s physiology by connecting millimeter-sizedmillimeter sized tissues—the beating heart muscle, the metabolizing liver, and the functioning skin and bone that are grown from the patient’s cells. [Our approach is] uniquely designed for studies of systemic conditions associated with injury or disease ... One patient at a time, from inflammation to cancer!”
Organoid intelligence: Johns Hopkins’Hopkin’s Thomas Hartung suggested that brain organoids have enormous potential as biocomputing systems. These systems promise to elucidate the biological basis of human cognition, improve disease modeling and drug testing, and complement traditional computing.
The presentation brought to mind a paper by Hartung’s team. Titled “Organoid intelligence (OI): the new frontier in biocomputing and intelligence-in-a-dish,” the paper noted that organoid intelligence relied on three-dimensional structures enriched with cells and genes associated with learning and memory; microfluidic perfusion systems capable of supporting scalable and durable culturing; organoid machine interfaces to capture spatiotemporal chemical and electrophysiological signaling; and AI-driven data analysis.
A Johns Hopkins team led by Thomas Hartung has outlined a plan to combine artificial intelligence and brain organoids to achieve “organoid intelligence.” The idea is to replicate critical molecular and cellular aspects of learning and memory and, possibly, aspects of cognition in vitro. Practical applications of organoid intelligence include disease modeling and drug screening.
According to the Hartung team, biocomputing systems, or wetware, could alleviate the energy-consumption demands of supercomputing, which are becoming increasingly unsustainable. In any case, biocomputing could complement traditional computing, given that human brains excel in processing complex information: “Brains perform both sequential and parallel processing (whereas computers can do only the former), and they outperform computers in decision-making on large, highly heterogeneous, and incomplete datasets.”
Biological structures other than brains may have much to tell us about cognition. This proposition animates much of the research led by Tufts’ Michael Levin, who extends the agential, purpose-driven view of cell collectives to what may be the most fundamental purpose: intelligence. Indeed, he has been developing concepts that he calls “basal cognition” and “diverse intelligence.”
“Basal cognition is the quest to understand how Mind scales—how large numbers of competent subunits can work together to become intelligences that expand the scale of their possible goals,” Levin writes. “Crucially, the remarkable trick of turning homeostatic, cell-level physiological competencies into large-scale behavioral intelligences is not limited to the electrical dynamics of the brain.”
According to Levin, a kind or degree of intelligence can be said to exist wherever biological entities navigate problem spaces, in keeping with philosopher Willam James’ definition of intelligence as “the ability to reach the same goal by different means.” For example, cell collectives of diverse types may show a kind of intelligence when they solve the problems of regulative embryogenesis, regeneration, and cancer suppression. For Levin, every intelligence is a collective intelligence because all cognitive systems, including those with “mere” basal cognition, are made from parts.
To test this idea, he has experimented with planaria and xenobots. For example, he found that in planaria that had been grown from headless (and brainless) fragments, memories about the location of food rewards—memories that had been obtained before decapitation—were retained. The observation has motivated studies of planarian regeneration that explore how “pattern memory,” for properties such as body shape, may depend on bioelectrical signaling.
Xenobots, clumps of skin cells that have been scraped from Xenopus frog embryos, can move toward a target, push a payload, self-heal after being cut, and display swarming behavior. After being incubated together, the cells collaborate to make the best of their new circumstances, assuming a living form that deviates from the form that one might have considered genomic destiny, and flapping their cilia in coordinated fashion. Xenobots can even self-replicate by collecting loose Xenopus skin cells. The behavior lasts for a few generations.
Levin and colleagues have also brought us Anthrobots, living robots that build themselves from human tracheal cells. They can assume different shapes, use their cilia to move about in various ways, and repair scratches in cultured human neural cell sheets.
In general, Levin’s work is about cracking the codes that determine how biological form and function emerge, codes that make use of genome-specified hardware while operating at levels above the genome. The work not only has implications for regenerative medicine, but also for the harnessing of the information processing, decision making, and cognitive capabilities behind physiology and morphogenesis.
To explore natural and synthetic morphogenesis, Levin and the other scientists cited in this article are looking beyond genomic reductionism. Instead, they are seeing where an appreciation of emergence might lead them. One possibility is a new understanding of intelligence.
Ordinarily, intelligence is seen as something that is concentrated in the human mind or a supercomputer and that regards nature at a remove, as a god might, but far more feebly. It cannot reduce all of nature to the locations and momenta of individual particles, and lacking these values for any given time, not to mention a god’s powers of calculation, it is unable to determine past or future values from the laws of mechanics. But perhaps intelligence can be seen as something that is both embedded in and distributed through living systems at every level, albeit in different degrees. As Levin noted in an editorial written with philosopher Daniel C. Dennett, it’s “cognition all the way down.”
Lorem ipsum dolor sit amet, consectetur adip elit. Donec posuere dolor massa, pellentesque aliquam nisl facilisis sed.
Fortunately, there is more than one kind of “down.” In science and engineering, the usual down involves a search for a system’s smallest and most fundamental components. But the dive for fundamental components, such as atoms and molecules, can descend past things that are fundamental in their own right, namely, the emergent properties of complex systems.
Emergent properties cannot be understood—in practice or even in principle—by identifying and characterizing individual parts. Instead, rules of interaction must be uncovered. But how? According to Nobel Prize–winning physicist Robert Laughlin, what’s needed is a bottom-down approach, which sounds paradoxical but really just refers to a different kind of down, one that gets to the bottom of what we might call interaction spaces.
In A Different Universe, Laughlin argued that physics needn’t focus too intently on fundamental particles. Instead, it can recognize the importance (or even primacy, in some contexts) of the interactions between simple components, and thus account for otherwise unexplainable emergent phenomena such as phase transitions and superconductivity. Something similar could be said of biology. It needn’t focus too much on genomic details. Instead, it can devote a large share of its attention to the interactions that pertain at higher levels of organization. For example, it can uncover the rules that explain how cells self-organize into tissues.
Sometimes, it is necessary to recognize when events at the level of cells or cell collectives are of central importance, despite the excitement over the possibilities offered by genomics. This view goes back as far as 2004, when microbiologist and biophysicist Carl R. Woese, most famous for defining the Archaea, offered his views on what “a new biology for a new century” would entail. Even though the Human Genome Project had been completed just a year earlier, Woese wrote that, for biology, “molecular paradigm [had] run its course,” and would be most suitable for engineering purposes. He added that if biology were to become society’s teacher and not be reduced to doing society’s bidding, it would need a more holistic paradigm, one that would consider factors such as “evolution, emergence, and biology’s innate complexity.”
Were Woese still with us, he might be surprised that an overly reductive molecular paradigm is becoming less and less satisfactory not just in biology proper, but in disciplines that blur the distinction between biology and engineering. For example, synthetic biology, a discipline devoted to improving the engineerability of living systems, mainly individual cells, is broadening its scope to encompass synthetic morphology, which seeks to control the forms and functions that can be assumed by cellular collectives.
As synthetic biology expands its remit from individual cells to cellular collectives, including tissues and organs, it is expanding its collection of tools. At present, most tools consist of genes and gene networks that are meant to give cells new abilities, such as the ability to consume unconventional feedstocks, secrete nonnatural compounds, and respond to physiological or environmental conditions (perhaps by serving as recording or signaling devices). But additional tools are being developed to influence how cells communicate with one another, typically via the modification of chemical, mechanical, and electrical signals.
Prospects for such tools were discussed in an early paper about synthetic morphology. The paper, written by the University of Edinburgh’s Jamie A. Davies, was published in 2008 and called for the creation of a library of sensor, regulatory, and effector modules. Connected functionally within cells, these modules could, Davies suggested, program cells to organize themselves into “specific, novel arrangements, structures, and tissues.”
Davies noted that the beginnings of such a library already existed. For example, he identified master regulators that could activate morphogenetic modules for 10 basic cellular events: apoptosis, cell proliferation, cell fusion, cell locomotion, chemotaxis, haptotaxis, cell–cell adhesion/condensation, cell sorting, epithelial-mesenchymal transition, and epithelial folding.
Davies and colleagues soon extended this work. In 2014, they reported that they had constructed inducible effector modules for the control of cell adhesion, elective cell death, cell fusion, cell locomotion, and proliferation. They also described how these modules were used in a human cell line to induce morphological behaviors on command.
“[The constructed modules] are produced in a standardized format, easy to connect to existing logic and sensory modules,” the researchers detailed. “We also report the testing of these synthetic, inducible modules in mammalian cells.” They added that they hoped to “extend engineered control of morphogenetic cell behavior to more complex 3D structures that can inform embryologists and may, in the future, be used in surgery and regenerative medicine, making synthetic morphology a powerful tool for developmental biology and tissue engineering.”
In the course of his research, Davies began referring to the “rules” of signal-driven behaviors, and to the cells and cell collectives that follow the rules as “agents.” For example, in a public lecture at the University of Edinburgh in 2010, Davies said that form and function emerged in kidneys through the action of “individual agents following simple rules in a dumb way to do something remarkable.”
Developing this idea further, Davies, along with Tufts University’s Michael Levin, described how synthetic morphology could be accomplished with agential materials. In a 2023 paper, they suggested that design strategies could use agential materials in addition to traditional gene- and protein-based engineering approaches. Essentially, the authors envisioned a collaborative approach, one in which designers cooperated with agential materials, which the authors recognized as having their own goals, agendas, and powers of problem-solving.
“Such an agential bioengineering approach could transform developmental biology, regenerative medicine, and robotics, building on frameworks that include active, computational, and agential matter,” Davies and Levin wrote. “We highlight experimental embryology studies … that go beyond the familiar, default outcomes of embryogenesis, revealing the plasticity, interoperability, and problem-solving capacities of life.”
Ideas like those expressed by Davies and Levin are being pursued by scientists at MIT’s Multi-Cellular Engineered Living Systems (M-CELS), which was launched in 2020. Whereas Davies and Levin refer to agential materials, M-CELS refer to “purpose-driven living systems with multiple interacting living components.”
M-CELS’s three founders—Roger D. Kamm, Linda G. Griffith, and Ron Weiss—all contributed to a 2018 article on the promise of multicellular engineered living systems. According to this article, M-CELS can be fabricated in at least two different ways: “top-down” assembly (such as building two- or three-dimensional forms and seeding or plating them with differentiated stem cells or primary cells) or “emergent engineering” (pluripotent cells subjected to guidance cues to drive differentiation and self-organization). The article also described M-CELS applications: organ-on-chip or tissue-chip systems (for disease modeling and drug discovery), implantable “hyper-organs” (for sensing biological signals and synthesizing and secreting a biologic product in response), and biological actuators or bio-robots (for drug delivery and microsurgery).
Progress toward such applications was evident at “M-CELS: Advances in Basic Research and Translational Opportunities,” an M-CELS-organized symposium that was held March 26–28, 2025, at Hilton Head, NC. At this symposium, most presentations described advances relevant to specific application areas, such as the following:
Vascularized organoids: The Wyss Institute’s Jennifer A. Lewis described how human induced pluripotent stem cells can be transformed into vascularized kidney organs via scalable differentiation protocols. She also reviewed her group’s recent work on embedding vasculature into human kidney tissue models via coaxial sacrificial writing in functional tissue (co-SWIFT), a bioprinting method capable of generating hierarchically branching, multilayered vascular networks within both granular hydrogel and densely cellular matrices. Earlier, in an announcement that accompanied the journal article describing this work, Lewis stated, “We were able to successfully 3D-print a model of the vasculature of the left coronary artery based on data from a real patient, which demonstrates the potential utility of co-SWIFT for creating patient-specific, vascularized human organs.”
Vascularized models were also discussed by UC Irvine’s Christopher C.W. Hughes. His laboratory has developed vascularized micro-organs emulating the liver, pancreas, lung, and brain (which includes a blood-brain barrier) as well as vascularized micro-tumors. The technology is currently licensed to Aracari Biosciences, which Hughes co-founded and now serves as CSO.
Innervated tissue models: Binghamton University’s Tracy Hookway discussed the engineering of innervated multicellular cardiovascular models, using pluripotent stem cells that have been differentiated into sympathetic and parasympathetic neurons. In Hookway’s work, neuronal cell populations are mixed with cardiomyocytes at different developmental stages to investigate their influence over beat rate, contraction force, and calcium-handling kinetics.
Immune organoids: Georgia Tech’s Ankur Singh discussed the development of synthetic, human ex vivo immune organoids to replicate the structure and function of immune tissues. In recent work, Singh’s group developed synthetic hydrogels mimicking the lymphoid tissue microenvironment, enabling human germinal centers from tonsils and peripheral blood mononuclear cell–derived B cells. In an announcement pertaining to this work, Singh stated, “Our synthetic hydrogels [allow] us to model antibody production from scratch, more precisely, and for a longer duration. This is a gamechanger for understanding and treating immune vulnerabilities in patients with lymphoma who have undergone cancer treatment—and hopefully other disorders too.”
Embryoid models: The University of Pittsburgh’s Mo Ebrahimkhani , a researcher attuned to the cellular cross-talk in advanced organoid models, described the heX-embryoid, an embryoid model that is derived from adult cells and shows self-organizing peri/post-implantation cellular programs, including the formation of the amniotic cavity and body axis generation. To date, the early post-implantation stages of human development have been difficult to study due to technical and ethical challenges. In press materials that accompanied heX-embryoid’s introduction, Ebrahimkhani asserted that the new technology would “unlock this ‘black box’ of human development, which could help solve the mystery of why about 60% of pregnancies fail in the first two weeks.”
Patient-specific multiorgan models: Columbia University’s Gordana Vunjak-Novakovic discussed her group’s development of organs-on-chip models of human pathophysiology. For example, she highlighted a multi-organ chip that incorporates different tissue niches that are separated by endothelial barriers, yet linked by a vascular flow containing circulating cells.
In a recent article, Vunjak-Novakovic and colleagues asserted that their models support communication between tissues while preserving their individual phenotypes. In a statement that accompanied the paper, she noted, “We can model a patient’s physiology by connecting millimeter-sizedmillimeter sized tissues—the beating heart muscle, the metabolizing liver, and the functioning skin and bone that are grown from the patient’s cells. [Our approach is] uniquely designed for studies of systemic conditions associated with injury or disease ... One patient at a time, from inflammation to cancer!”
Organoid intelligence: Johns Hopkins’Hopkin’s Thomas Hartung suggested that brain organoids have enormous potential as biocomputing systems. These systems promise to elucidate the biological basis of human cognition, improve disease modeling and drug testing, and complement traditional computing.
The presentation brought to mind a paper by Hartung’s team. Titled “Organoid intelligence (OI): the new frontier in biocomputing and intelligence-in-a-dish,” the paper noted that organoid intelligence relied on three-dimensional structures enriched with cells and genes associated with learning and memory; microfluidic perfusion systems capable of supporting scalable and durable culturing; organoid machine interfaces to capture spatiotemporal chemical and electrophysiological signaling; and AI-driven data analysis.
A Johns Hopkins team led by Thomas Hartung has outlined a plan to combine artificial intelligence and brain organoids to achieve “organoid intelligence.” The idea is to replicate critical molecular and cellular aspects of learning and memory and, possibly, aspects of cognition in vitro. Practical applications of organoid intelligence include disease modeling and drug screening.
According to the Hartung team, biocomputing systems, or wetware, could alleviate the energy-consumption demands of supercomputing, which are becoming increasingly unsustainable. In any case, biocomputing could complement traditional computing, given that human brains excel in processing complex information: “Brains perform both sequential and parallel processing (whereas computers can do only the former), and they outperform computers in decision-making on large, highly heterogeneous, and incomplete datasets.”
Biological structures other than brains may have much to tell us about cognition. This proposition animates much of the research led by Tufts’ Michael Levin, who extends the agential, purpose-driven view of cell collectives to what may be the most fundamental purpose: intelligence. Indeed, he has been developing concepts that he calls “basal cognition” and “diverse intelligence.”
“Basal cognition is the quest to understand how Mind scales—how large numbers of competent subunits can work together to become intelligences that expand the scale of their possible goals,” Levin writes. “Crucially, the remarkable trick of turning homeostatic, cell-level physiological competencies into large-scale behavioral intelligences is not limited to the electrical dynamics of the brain.”
According to Levin, a kind or degree of intelligence can be said to exist wherever biological entities navigate problem spaces, in keeping with philosopher Willam James’ definition of intelligence as “the ability to reach the same goal by different means.” For example, cell collectives of diverse types may show a kind of intelligence when they solve the problems of regulative embryogenesis, regeneration, and cancer suppression. For Levin, every intelligence is a collective intelligence because all cognitive systems, including those with “mere” basal cognition, are made from parts.
To test this idea, he has experimented with planaria and xenobots. For example, he found that in planaria that had been grown from headless (and brainless) fragments, memories about the location of food rewards—memories that had been obtained before decapitation—were retained. The observation has motivated studies of planarian regeneration that explore how “pattern memory,” for properties such as body shape, may depend on bioelectrical signaling.
Xenobots, clumps of skin cells that have been scraped from Xenopus frog embryos, can move toward a target, push a payload, self-heal after being cut, and display swarming behavior. After being incubated together, the cells collaborate to make the best of their new circumstances, assuming a living form that deviates from the form that one might have considered genomic destiny, and flapping their cilia in coordinated fashion. Xenobots can even self-replicate by collecting loose Xenopus skin cells. The behavior lasts for a few generations.
Levin and colleagues have also brought us Anthrobots, living robots that build themselves from human tracheal cells. They can assume different shapes, use their cilia to move about in various ways, and repair scratches in cultured human neural cell sheets.
In general, Levin’s work is about cracking the codes that determine how biological form and function emerge, codes that make use of genome-specified hardware while operating at levels above the genome. The work not only has implications for regenerative medicine, but also for the harnessing of the information processing, decision making, and cognitive capabilities behind physiology and morphogenesis.
To explore natural and synthetic morphogenesis, Levin and the other scientists cited in this article are looking beyond genomic reductionism. Instead, they are seeing where an appreciation of emergence might lead them. One possibility is a new understanding of intelligence.
Ordinarily, intelligence is seen as something that is concentrated in the human mind or a supercomputer and that regards nature at a remove, as a god might, but far more feebly. It cannot reduce all of nature to the locations and momenta of individual particles, and lacking these values for any given time, not to mention a god’s powers of calculation, it is unable to determine past or future values from the laws of mechanics. But perhaps intelligence can be seen as something that is both embedded in and distributed through living systems at every level, albeit in different degrees. As Levin noted in an editorial written with philosopher Daniel C. Dennett, it’s “cognition all the way down.”
Lorem ipsum dolor sit amet, consectetur adip elit. Donec posuere dolor massa, pellentesque aliquam nisl facilisis sed.
.svg)
.gif)