A conversation with Christof Koch about his new book and theories of consciousness
Christof Koch is a remarkable kind of scientist: one who has been at the top of his field for decades, is considered a “hard-nosed” experimental biophysicist, neuroscientist and electrophysiologist, but also a scientist who has been willing to consider the more humanistic side of consciousness in his work.
His most recent book, The Feeling of Life Itself: Why Consciousness is Widespread But Can’t Be Computed, attempts to bridge the gap between these scientific and humanistic tendencies.
It is both an overview of the problem — how can we explain consciousness in ourselves and in non-human animals, and possibly further down the chain — and a narrative about Koch’s preferred set of solutions that revolve around the Integrated Information Theory of consciousness (IIT). Koch’s book also includes some reflections on mystical experience and ethics, without making those additions feel like slapdash add-ons.
IIT is a now-prominent theory of consciousness that has been developed primarily by neuroscientist Giulio Tononi and his colleagues, with significant support in recent years from Koch himself. Much ink has been spilled about IIT, including Tononi’s own 2012 book, Phi: A Voyage From the Brain to the Soul, but Koch’s new book includes a relatively accessible description of the problems that IIT seeks to answer and also how it answers them. It’s a mostly non-technical work, but some chapters require a little bit of technical slogging. It’s worth it.
We are at the point in the development of theories of consciousness that empirical tests are being developed to distinguish various theories. The Templeton Foundation is funding a multi-year effort to distinguish the predictions of IIT and another prominent theory: Dehaene and Changeux’s Global Neuronal Workspace theory of consciousness.
Koch’s book discusses this rivalry in polite terms but offers many reasons why he thinks IIT is the more empirically-accurate and predictive theory of consciousness. It feels at times like a preemptive strike in the “official” faceoff that is underway between these two theories, particularly in the debate over whether the primary “neuronal correlates of consciousness” are in the posterior “hot zone” of the brain, as Koch argues, or more in the prefrontal cortex area of the brain, as Dehaene and Changeux argue.
I’ve known Koch for almost a decade. I interviewed him for a documentary film project many years ago, conducted a pair of in-depth written interviews (with this one being the second), and persuaded him to write the foreword to my 2014 book of essays centered around panpsychism, Eco, Ego, Eros: Essays in Philosophy, Spirituality, and Science.
In the below interview, Koch and I discuss his new book, his latest thinking on the nature of consciousness and computation, and also some differences between his and Tononi’s IIT approach and the General Resonance Theory of consciousness that I and Jonathan Schooler have developed in recent years.
Koch has been a working neuroscientist and biophysicist for three decades, and was a full professor at CalTech until 2013, at which time he moved to the Allen Institute for Brain Science. He is currently president and chief scientist at the Allen Institute.
Tam — What tips you over the threshold in deciding to write a new book? You’ve written three now, with your first focused on the neuroscientific quest for the physical basis of consciousness, your second taking a more biographical and “romantic” twist on similar themes, and your latest tackling again similar themes but adding half of the book describing in some detail the Integrated Information Theory of consciousness.
Christof — The study of consciousness and its footprints in the brain is in a very active phase worldwide. Given the rapid evolution of recording technologies — from Neuropixels electrodes that can measure the spiking activities of thousands of cells at sub-millisecond resolution; to newer, much cheaper and easier to operate non-invasive brain imaging technologies that can pick up subtle signals in the human brain at the mm-msec scale; and of causal tools that allows us to intervene and to manipulate molecularly identified neuronal populations transiently, deliberately, reversibly and in a sustained manner — the field is exploding with new experiments and data that need to be analyzed and interpreted in light of the various extant theories of consciousness.
Given all this progress and the rapid development of what I consider an empirically-testable, consistent, fundamental theory of consciousness — Integrated Information Theory (IIT) — I felt the time was ripe to update my previous two books.
While Confessions of a Romantic Reductionist was partially driven by a personal need to explicate my philosophical positions vis-à-vis the Big Questions of Life, The Feeling of Life Itself is a more straightforward scientific and intellectual account of what we know about the physical substrate of consciousness in the human brain, IIT, building a consciousness-meter, how to test IIT and its philosophical and ethical implications.
Tam — Does your and Crick’s “neurobiological theory of consciousness,” which you fleshed out in a number of papers and in your 2004 book, The Quest for Consciousness, mesh naturally with the Integrated Information Theory of consciousness? Or have you abandoned or modified some aspects of your own earlier theories in your adoption of IIT in your own work?
Christof — Crick’s and my quest for the neuronal footprints of consciousness was partially galvanized by the (re)-discovery of synchronized discharges in the visual cortex of cats and monkeys looking at moving bars and other visual stimuli in the late 1980s. Scientists discovered that neurons fired action potentials not haphazardly but in a periodic manner, with a pronounced tendency for action potentials (spikes) to be spaced 20–30 milliseconds apart. These waves in the recorded electrical potential are known as gamma oscillations, roughly centered around 40 Hz. Even more remarkably, nearby neurons that signal the same object fired action potentials roughly simultaneously. This led us to propose that these 40 Hz oscillations are a neuronal correlate of consciousness (NCC).
This simple idea captured the attention of scientists and helped power the modern quest for the NCC. Yet after more than a quarter of a century of empirical investigations of 40 Hz oscillations in people and animals in hundreds of experiments, the verdict is that they are not a true NCC! What emerged is a more nuanced view of the relationship between gamma oscillations and consciousness. Periodic firing of neurons in this range of gamma band activity in the EEG is closely linked to selective attention, with synchronized gamma oscillations between two regions strengthening the effective connectivity within the two underlying neuronal coalitions. Just as we can pay attention to something that we don’t see, stimuli can trigger 40 Hz oscillations without evoking any conscious experience. That is, gamma synchrony can occur without consciousness.
Note that Crick and I never thought of our 40 Hz oscillation hypothesis as a “theory” in any rigorous sense but as a working hypothesis, a shrewd guess that gamma oscillations are a hallmark or trait of the NCC. Even if this hypothesis does turn out to be correct it is not a theory in itself.
Tam — With the modification of IIT in its 3.0 version to redefine consciousness as the maximally-irreducible cause-effect power (“MICE”) of a given network, would it be more accurate to think of the theory as being about integrated causal powers than integrated information per se?
Christof — Correct. According to the theory, ultimately there is nothing in the universe but causal powers, both extrinsic ones that physics and allied sciences study, and intrinsic powers that are consciousness. As the ancient Epicureans would have said, there is nothing but causal powers and the void!
The mathematical measure called integrated information, quantified by the non-negative number phi (the associated Greek symbol is written as Φ), is a measure of the maximal intrinsic causal power of any mechanism in a state. The bigger Φ, the more the system exists for itself, the more it is irreducible, the more it is conscious. A system with zero integrated information, or zero Φ, does not feel like anything. It is not conscious. It does not exist for itself, as it can be fully reduced to its subcomponents.
IIT is an axiomatic theory in the sense that it starts with five fundamental properties of any conscious experience, proceeds to five associated postulates concerning the physical substrate of any physical mechanism, evolved or engineered, that is conscious. IIT does not require any further hypothesis about neuronal or cognitive processes, parameter fudging or ad hoc assumptions. In that sense, it is complete, consistent, general and substrate-independent.
Of course, any scientific theory must ultimately be judged not for its elegance but for the purview and precision of its explanatory and predictive powers.
Tam — You discuss mystical experiences in your book but you don’t dwell much on how IIT would reflect mystical experiences. Are these states instances of higher integration, lower integration, or just different types of integration across different brain circuits?
Christof — There is no question that mystical experience, whether induced by religious practices, meditation, the ingestion of entheogenic psychoactive substances (such as DMT, 5-methoxy-DMT, ayahuasca and so on) or by near-death experiences, are highly unusual and exceptional. William James, the founder of American psychology, characterized mystical experiences in his Varieties of Religious Experience as being less of an emotional state (although they are very intensely felt) but more of an intellectual state — the subject deeply understands something about the universe and their relationship to it, although they are unable to put this knowledge into words. James refers to their ineffable and noetic quality. Even though mystic experiences are short-lasting, their effects can liberate enormous amounts of energy and enthusiasm and can upend and transform the life of the experiencer (e.g., the Pauline conversion on the road to Damascus or Blaise Pascal experience).
In such mystical experience, the ego disappears — no self, no body, no dreams, fears, desires or hopes. Furthermore, space and time also disappear as does the external world. What remains is a disembodied non-self with an intense experience of the sublime, of terror and/or ecstasy, the burning furnace of being. Remember that the etymology of awful is “worthy of respect or fear, striking with awe”.
This comes close to a state of pure experience (also known as pure or naked awareness) that Tibetan Buddhists strive for, or that can possibly be experienced in a floatation tank — suspended in a sightless, soundless, odorless, bodiless, timeless, egoless, and mindless space. If these are truly little differentiated (with attendant low Φ) but of high consciousness in a phenomenal, subjective sense, this would be a challenge to IIT.
Language notoriously fails at describing these mystical states (thus, their ineffable nature). It is challenging from a third-person point of view to understand what it means for the very rich phenomenology of space, even of a blank and empty canvas, to disappear during a mystical experience (unlike, say, when I wake up in the complete dark yet have a sense of visual space that is not filled with any visible object). Likewise, how can an experience be timeless? But many people do describe mystical experiences as being out of time, without feeling that the experience is too brief, or too long, or of wanting it to end.
At this point in time, the exact phenomenological character of mystical experiences and their degree of differentiation remain open research questions that I seek answers to. Let me end on an essential point that James used in 1902 to conclude his investigations -
“Yet, I repeat once more, the existence of mystical states absolutely overthrows the pretension of non-mystical states to be the sole and ultimate dictators of what we may believe.”
Tam — You mention a number of times in your 2004 book resonance and standing waves as a potential key for understanding the nature of consciousness, but you don’t mention these terms at all in your latest book. For example, you wrote in your 2004 book, Quest:
If the input [to the brain] is more sustained and is boosted by top-down attention … a sort of standing wave or resonance might be created in the network, with vital contributions from the feedback pathways. Both local and more global feedback could cause neurons to synchronize their spiking activity above and beyond the degree of synchronization that results from the sensory input by itself. This increases their postsynaptic punch compared to when they fire independently. A powerful coalition of neurons could be assembled in this manner, able to project its influence to the far reaches of the cortex and below.
Has your thinking changed on this particular issue or just your vocabulary? For example, p. 142 of your new book uses the terms “sustained feedback,” “recurrent” and “reentrant” processing in looking at similar ideas, but not “resonance” or “standing waves.”
Christof — The experimental evidence for such resonances or standing waves is ambiguous. At the Allen Institute for Brain Science that I lead, we have just submitted a paper to bioRxiv (Siegle et al. “A survey of spiking activity reveals a functional hierarchy of mouse corticothalamic visual areas” 2019) in which we recorded the spiking activity of close to 100,000 neurons in the visual system of awake mice. Given this vast amount of data, bigger than anyone has ever previously recorded (all of which is freely and publicly available for anyone to analyze), it is easy to find all sorts of interesting dynamic phenomena, including various oscillations that wax and wane within overlapping spectra. Yet so far, no obvious relationship to behavior or to function has emerged.
Note that the vast majority of psychophysical or cognitive experiments confound selective visual attention and visual consciousness. When the effects of conscious visibility are properly distinguished from those of selective attention, high gamma synchronization is tied to attention, independently of whether or not the stimulus was seen by the subject, while mid-range gamma synchronization relates to visibility (Aru et al., 2012; Wyart & Tallon-Baudry, 2008). Hermes et al. (2015) demonstrated the existence of gamma oscillations in human visual cortex but only when looking at specific types of images. This led to the conclusion that gamma-band oscillations are not necessary for seeing (Ray & Maunsell, 2011). Finally, gamma synchrony can persist or even increase in early NREM sleep, during anesthesia (Imas et al., 2005; Murphy et al., 2011) or seizures (Pockett & Holmes, 2009) and can be present for unconscious emotional stimuli (Luo et al., 2009). That is, gamma synchrony can occur without consciousness.
The status of the current evidence is messy. Of course, all the tools at the disposal of us neuroscientists are edentate when confronted by the vast tapestry of 16 billion heavily networked nerve cells that make up the human cortex, the most complex piece of highly active tissue in the known universe.
Tam — This is very interesting data, highlighting well the complexity of these issues! While the data about gamma synchrony indeed seems ambiguous at this juncture, what about the notion of resonance or standing waves more generally as a general principle for the organization of consciousness? Is this still part of your thinking?
Christof — I know that your thinking about consciousness in the brain is heavily influenced by resonances at particular frequencies (gamma, beta, theta and so on; see below). While at this early stage of the exploration of the brain it would be foolish to categorically rule out any physical process, as an electrophysiologist I’m less enthused about ascribing specific functions to specific frequency bands, let alone experience for two reasons.
Firstly, by and large, the causal actors between neurons that act at the time scale relevant for consciousness (5–500 msec) are action potentials that cause, in turn, synaptic release of packets of neurotransmitters. Most neurons fire highly irregular spike trains, more compatible with a random Poisson process than with a highly synchronized, clocked process of the sort we’re familiar with from electronic circuits. Yes, in a deeply asleep cortex, neuronal on-off states occur with a high degree of regularity every couple of 100 msec, leading to theta band oscillations. Furthermore, given the widespread feedback nature of excitatory pyramids cells and inhibitory interneurons, certain frequencies — such as in a broadly defined gamma band extending from 30 to perhaps 100 or more Hz — can be commonly found in the awake cortex. So yes, the EEG that is recorded from the scalp outside the skull and its sibling, the local field potential (LFP) that is recorded with thin electrodes inserted into cortex proper (through the skull), all show peaks at particular frequencies. Yet these are broad and are superimposed onto a 1/f^n type of power-law decay characteristic of many natural systems (see the figure).
Secondly, the extent to which oscillations in the LFP or the EEG have causal influence on firing pattern of neurons remains an open question. Consider the sounds the beating heart makes. These can be picked up by a stethoscope and can be used to diagnose cardiac conditions. However, there is no evidence that the body exploits these sounds for any function.
My own group has provided some electrophysiological in vitro evidence that oscillations in the extracellular field at particular frequencies may be able to entrain spikes in a cell-type dependent manner (Anastassiou, Perin, Markram & Koch Nature Neuroscience 2011). At this point, we do not know what role such so-called ephaptic coupling (to distinguish them from the more conventional synaptic coupling) play in the human brain.
Tam — As you know, I’ve developed a theory of consciousness (General Resonance Theory of consciousness), with Jonathan Schooler at UC Santa Barbara, that has much in common with IIT but also some key differences. Without getting too much into the weeds, do you see any merit in the arguments posed by various authors (including in my 2014 paper, “Taking time seriously in Integrated Information Theory”) that the exclusion principle of IIT has some issues, including that it feels ad hoc and arbitrary? Have you personally entertained alternatives to the exclusion principle?
Christof — Personally, I have not entertained any alternatives as I see no need for them. The exclusion postulate is derived from the indubitable assertion that any experience is definite. Whatever I experience is in my consciousness at that point in time and the rest, what I don’t experience, is not part of that experience. That is, my experience has definite borders, including a small number of distinctions, and excluding the vast majority.
The world I see isn’t bordered by a line beyond which things are grey or dark, such as behind my head. Rather, my experience simply doesn’t extend outside the liminal.
Like a canvas — the strokes of the brush are painted onto the canvas, everything else is not. Even when I see something fuzzy, like through a fog, then that will be the character of my experience.
This is even true of the mystical experiences that we spoke about. The timelessness, space-lessness, oceanic boundlessness, and being in complete unity with the universe reported by mystics and psychonauts defines that particular experience, neither less nor more. I simply can’t conceive of a world in which experience isn’t definite.
This is what the exclusion principle expresses in a particularly elegant manner (elegant to me as a physicist) about the physical substrate of any conscious experience. As you express in your article, the fact that (per IIT) only the maximum of intrinsic causal power exists from an intrinsic point of view appears strange to many. However, there are any number of extrema principles in physics in which either only the minimum or the maximum of some function is considered.
Take the principle of least action, a key theme in relativity theory, thermodynamics, mechanics, and fluid mechanics. It stipulates that of all the ways that a particular physical system can evolve, only one actually occurs, the one that is an extremum. For instance, the shape of a sagging bicycle chain or a metal ring chain suspended at both ends is the one that minimizes its potential energy
I find the exclusion principle compelling at the epistemological level as I have no evidence what-so-ever of a group mind. The exclusion principle rules them out. It’s a very powerful principle that does have, as you and Schwitzgebel independently point out, counter-intuitive consequences. The one I discussed at length in my last book is brain-bridging between, say, your and my brain — creating a sort of artificial corpus callosum (like the 200 million fibers connecting the left and right cortical hemispheres) but now directly linking, neuron to neuron, our two brains. If we had a technology that could do this, we could reach a critical threshold of such fibers where adding a single additional wire to the brain-bridge linking our two nervous systems — one axon out of millions — would turn our two disparate minds (each one of which accesses some part of the other person’s brain) into a new conscious mind. You and I would suddenly disappear — per the Exclusion principle — and be replaced by a single, conscious Übermind consisting of some combination of you and me.
This single mind would see through four eyes, hear with four ears and so on, and would control both bodies. If this single axon was then removed again, the integrated information of our separate minds, each one housed in its own skull, would (in this thought experiment) once again exceed the integrated information of both brains combined, the Übermind would disappear, and your and my mind would once again exist independently. Thus, adding and removing this one axon would flicker between two conscious minds and a single Übermind. This possibility could, in principle, be empirically tested.
Tam — I certainly agree that mystical experiences, like any human experiences, should be considered relevant data for our theories of consciousness. But arguing that mystical experiences are timeless strikes me as more poetry than actuality (and I’ve had my share of deep mystical experiences, both spontaneous and induced in various sacred settings). Doesn’t for anything “to occur” or “to happen” require, by definition, occurrence in time, happening in time, in terms of there being a necessary passage of time for literally every experience? Similarly, if we measure the brain activity of meditating subjects or those undergoing psychedelic experiences, there is always complex brain activity of various types. Without a happening in time what would it mean for such a “timeless” experience to have occurred? Similarly, recognizing that vibrations, waves, resonance, are core to consciousness, as you do in much of your work, seems to necessarily imply a passage of time at the root of consciousness.
Christof — I can tell you on complete authority, that of my own, personal consciousness, that experiences can be timeless. I experienced the sublime, the terror and ecstasy, the awfulness of pure experience lasting for an eternity — without any feeling that this experience was too brief or too long, or of wanting it to end. And this is also reported by mystics of all cultures and ages.
You are correct, of course, in that psychological states — of terror, ecstasy and so on — inferred from a third person point of view always take place in time. So, one can talk of the time before my mystical time-less experience, and the time afterwards. Furthermore, the underlying physical substrate will (of course) evolve in time, with a before and an after in the brain-imaging experiments you refer to. But the experience itself can certainly be timeless, without any passage of time, outside of time.
Tam — If the core of IIT is that the whole is more than the sum of its parts (this is the definition of “integration”), is there a possible contradiction between this notion and the exclusion principle that all subsidiary conscious entities are extinguished? In other words, if it is the additional causal connections between constituent parts that comprise phi and thus the higher-level conscious entity, why isn’t there “leftover” causal power sufficient for the constituent entities to maintain their own level of consciousness while still being part of a larger whole?
Christof — Because it is not a maximum. Only the maximum of intrinsic causal power has intrinsic existence, or feels like something.
Tam — You write in your new book that electrical activity in the neocortex “is the physical substrate of experience” (p. 59). How literally do you take this statement? Is consciousness literally electrical activity, comprised of the electromagnetic fields that we measure with EEG, MEG, etc.? If so, have you surveyed the various electrical field theories of consciousness like Susan Pockett’s, Johnjoe McFadden’s, Mostyn Jones’, etc.?
Christof — Again, assuming IIT is correct — which is, after all, an empirical question — it answers this question precisely. There will be a level of granularity (molecules, organelles, synapses, single neurons, groups of neurons, microcolumns and so on) and spatial- and temporal-grain at which intrinsic causal power will be maximal. That is the level that defines that physical substrate of consciousness. Again, this can be tested and we have described such tests (Tononi, Boly, Massimini & Koch Nature Neuroscience 2016).
Given our very imperfect contemporary knowledge of the nervous system, what level is this likely to be? As a practicing neuroscientist, my intuition (which could be wrong, of course) is that this level is that of groups of molecularly-characterized cortical cells types and their electrical activity. Not the level of the extracellular field potential, whose low-frequency components are also known as the local field potential (LFP).
Why? Simply because, as described above, the causal power of the LFP onto the firing activity of neurons is weak. I have actually carried out such measurements (Anastassiou, Perin, Markram & Koch Nature Neuroscience 2011). We are doing further such tests to see whether the LFP can help synchronize groups of firing cells — if such a mechanism is present, it is likely to be weak (given the strength of the electrical field under physiological conditions which is in the 1–10 mV/mm range; see the review Buzsaki, Anastassiou & Koch Nature Rev. Neurosci. 2012).
Given that the extracellular field potential/LFP is the consequence of the current flow across the membranes of millions of synapses, dendrites and axonal segments in a resistive cytoplasm within 10–100 micrometers of any point in the brain tissue, it probably does not have the representational bandwidth to encode the highly specific content of any one conscious experience. But groups of neurons do.
Tam — how are conscious experiences encoded in groups of neurons?
Christof — by the physical substrate in the brain that constitutes, at this particular point in time, a maximum of intrinsic cause-effect power. This can, in principle, be measured. Ongoing experiments are trying to locate this in the posterior hot zone of the neocortex.
Tam — Similarly, could it be the case that the NCC are in fact particular electrical field patterns and not any particular neurological structures? Dehaene’s focus on P3 waves and long-range electrical connections across cortex seems to adopt this approach.
Christof — Well, the P3b evoked potential could simply be a read-out, just like the sounds that the heart makes when it beats is a highly useful diagnostic marker for the trained cardiologist without having any causal action on the heart or the rest of the body. The bottom line is that my knowledge of the biophysics of the brain makes me skeptical that the extracellular field potential/LFP is an important causal actor in consciousness.
Tam — drilling down further on the potential causal role of EM fields (LFP) in consciousness, McFadden makes the case in a 2013 paper that there is some good evidence for the brain’s EM fields to exert causal power back on the brain, suggesting that there is a dance back and forth between the biophysical structures of the brain and the EM fields produced by the brain — as your own work has found to some degree (as the papers you cite show). McFadden argues that recent experimental data demonstrates “that synchronous neuronal firing does indeed have a functional role in the brain; and also, that brain’s endogenous EM field is involved in recruiting neurons to synchronously firing networks.” Chiang, et al. made some waves in 2019 with a new paper finding further good evidence for non-synaptic ephaptic field effects on neuron function (which you mention above), stating: “Results support the hypothesis that endogenous electric fields, previously thought to be too small to trigger neural activity, play a significant role in the self-propagation of slow periodic activity in the hippocampus.” So, if this active area of research continues to yield good data showing EM field impacts on neuron function, could we see a new consensus form around the possibility that the EM field is the seat of consciousness?
Christof — As an experimentalist, I am skeptical of these claims, in particular given their statistical validity and effect size. Of course, at this point, no neuronal mechanisms, can be definitely ruled out (including exotic macroscopic quantum effects), as long as they don’t violate the laws of physics.
Tam — Turning to one of the core contentions of your book, that computers using von Neumann architectures, could never be truly conscious, does the growing development (e.g., Diaz-Alvarez et al. 2019) of neuromorphic-architecture computers, which rely in some cases on “self-assembly,” change this conclusion? That is, if computers rely on self-assembly and reentrant structures rather than feedforward networks (von Neumann architecture), do you see computers possibly becoming conscious in a meaningful manner in the coming decades?
Christof — Yes, you are entirely correct. There is nothing super-natural about the human brain. If the intrinsic causal powers of the brain are reproduced in an artificial, engineered substrate, then it too would have conscious experiences. This might happen in dense neuromorphic architectures or possibly in highly entangled quantum computers, such as the ones Google is building in Santa Barbara.
Tam — Last, you end your book with ethical considerations that flow from panpsychism and IIT. You suggest that we should expand our regard for other animals of all types because they all have the capacity to feel pain and to suffer. I share this view and when I first delved into the literature around panpsychism I found that it was not only intellectually satisfying as a theory of mind but also quite satisfying as the basis for an expanded ethics and environmentalism. Do these considerations lead you to conclude that veganism or vegetarianism flows naturally from this philosophical/scientific framework?
Christof — Yes, IIT shares many intuitions with biopsychism, the belief that all life, to a smaller and larger extent, is sentient, capable of experience, in particular of pain and pleasure. Thus, they too have a right to exist. Of course, such an ethical demand clashes with the innate desire of every creature to maintain itself. This requires metabolic energy that comes from ingesting other living creatures. All ethical and moral judgements always involve a balancing act of various rights.
I am primarily concerned with the capacity of large-brained creatures such as mammals, birds, fish, cephalopods, and so on, to suffer. Minimizing their suffering naturally leads to vegetarianism or, even better but more challenging, to veganism.
Another justification for why we should stop raising and killing hundreds of millions of animals for their flesh, milk, eggs, skin and so on, is the very high environmental footprint associated with animal husbandry. I am very encouraged by trends in the developed world toward vegetarianism and the growing importance of nutritious and tasty alternative foods, such as Impossible Foods and Beyond Meat as well as the wide-spread adoption of non-dairy milk and yogurts. We are moving, very slowly, in the right direction.
