What scientists know today about the functional architecture of the human brain
For centuries – going back to Descartes and in some respects as far as Aristotle and Plato – philosophers have wrestled with the ‘mind-body problem,’ the question of how a physical organism can give rise to, or somehow accompany, the astounding complexity of human thoughts, feelings, and experiences.  Today it seems that a historic breakthrough may be in the offing, because neuroscientists are making rapid progress toward mapping the functional architecture of the human brain. It seems likely that over coming decades some of the most profound mysteries of the brain may finally come to be penetrated.
What will these discoveries mean for us? Will we humans be compelled by the findings of neuroscience to see ourselves as ‘mere’ machines made of organic matter – extremely complex machines, to be sure, but machines nonetheless? Or is there something fundamentally misleading about using a machine metaphor for the operations that characterize the brain’s functioning?
From a practical standpoint, moreover, will our newfound knowledge allow us to reverse-engineer the human brain? Will we develop technologies for manipulating the neural processes that underlie our own thoughts, memories, and emotions? What happens, in such a world, to the deepest qualities that make us human?
In this Appendix I first give a brief summary of the current state of scientific knowledge about the human brain, then lay out some of the key areas in which scientists still lack a solid understanding of the brain’s functioning.
* * *
I. How much do scientists know today about the brain’s functioning?
1. Neurons and synapses.
Scientists have made great progress in identifying the electrochemical functioning of single neurons and small networks of neurons.  They understand quite well the physiology of synaptic transmission, and they have confidence that all the important operations of the brain, without exception, are grounded in the interactions of neurons. In particular, they have characterized in detail the electrochemical basis of long-term potentiation and synaptic plasticity, which they believe constitute the key underlying mechanisms, at the cellular level, of memory, learning, and ultimately of all the higher functions of cognition and mental experience.
2. Modular organization with anatomical regions specialized according to function.
Starting with the remarkable efforts of the German biologist Korbinian Brodmann more than a hundred years ago, scientists have mapped with increasing accuracy the roles played by different brain regions in generating particular aspects of mental function.  The processing of visual sensory data, for example, follows a specific pathway from the retina of the eye down the optic nerve, to the lateral geniculate nucleus in the thalamus, to the primary visual processing center at the back of the brain in the occipital lobe. Scientists warn us, however, not to interpret this spatial “division of labor” in an excessively rigid or literal-minded way: they underscore the fact that most key brain functions are distributed across many brain regions, and arise from the interactions and feedback-loops that link various brain regions in a continuously self-reflexive network of electrical signaling and processing.
Having said this, however, scientists have identified several dozen major brain regions that are involved as key players in generating particular aspects of mental function: from physical coordination (cerebellum) to the experience of emotion (amygdala); from speech recognition (Broca’s area) to the processing of experiences into long-term memories (hippocampus). Each mental function is linked to a specific region of the brain that constitutes its primary (though rarely its sole) locus of generative activity.
3. Inter-modular pathways and the “reverberating network” of feedback loops.
Most brain regions possess two-way connections with other brain regions that function in tandem with them. In some cases these interlinked regions are spatially contiguous with each other, while in other cases they are relatively far-removed in separate areas of the cranium. Thus, each brain region is continuously engaged in a process of feedback signaling, or two-way causation: it sends signals that modify the activity in a target region somewhere else, and the target region sends signals back to the originating region, modifying that originating region’s activity in turn.
Furthermore, most brain regions possess more than one such relationship of feedback linkage with another region. Scientists are therefore faced with the daunting task of tracing the causal interactions among multiple brain modules that are all networked with each other, and that are all simultaneously modifying one another’s activity in a continuous, multi-centered, and multidirectional web of signaling.
Consider the following example, drawn from among the hundreds of brain subsystems that neuroscientists have been mapping in recent years: the circuits that control motor movement.  These circuits are known to involve at least the following sets of neural connections between the motor cortex (near the top and middle of the brain in the frontal lobe) and various basal nuclei deep in the brain’s interior core:
a. Motor cortex to caudate nucleus
b. Caudate nucleus to internal globus pallidus
c. Internal globus pallidus to external globus pallidus
d. External globus pallidus to thalamus
e. Thalamus back to motor cortex (indirect feedback loop)
At the same time, additional signaling is also taking place directly and indirectly among various subsets of these same players:
a. Caudate nucleus to putamen
b. Putamen to external globus pallidus
c. External globus pallidus to thalamus
d. Thalamus back to caudate nucleus (indirect feedback loop)
a. Putamen back to caudate nucleus (direct feedback loop)
b. Internal globus pallidus back to caudate nucleus (direct feedback loop)
a. Internal globus pallidus to substantia nigra
b. Substantia nigra to caudate nucleus (indirect feedback loop)
Within this neural network, we can see that one particular site, the caudate nucleus, plays a major role in the control of motor movements – but it is far from clear exactly what that role is, and how one might go about disentangling the complex set of causal interconnections through which it plays its part. The caudate nucleus is connected in direct or indirect feedback loops with at least six other elements in this network: the motor cortex, the internal globus pallidus, the external globus pallidus, the putamen, the thalamus, and the substantia nigra. At the same time, all those other elements, while they interact with the caudate nucleus, are simultaneously signaling with each other in direct or indirect ways that further modify each other’s activity – and this in turn simultaneously alters their activity vis-à-vis the caudate nucleus.
This type of organizational structure, characterized by criss-crossing inter-modular pathways that form a layered network of feedback loops, appears to be extremely common in the functional architecture of the brain.
4. Overall organization
The brain’s two hemispheres are symmetrically arranged, with most anatomical features and functional elements duplicated on both the left and right sides. The two hemispheres are by no means identical, however, and some functions, such as language processing, are located primarily on one side (in this case the left hemisphere).
One way to break down the brain’s daunting complexity is by conceptualizing it as possessing a vertical arrangement into four major physical and functional layers: 
– Cortical regions, the brain’s uppermost and outermost areas: conscious sensation, abstract thought processes, planning, working memory.
– Limbic regions, tucked away below the cortex and around the brain’s inner core: emotional and instinctive behaviors and actions, as well as long-term memory.
– Midbrain, the brain’s inner core, perched on top of the brainstem: pre-processing center for sensory and other information; relays signals between brainstem below and cortex above.
– Brainstem, located below the brain’s core, atop the uppermost part of the spinal column: handles autonomic control mechanisms for metabolism, attention, and circadian rhythms.
This vertical description, however, can easily mislead us because it fails to capture the profoundly distributed and interactive nature of most brain functions: all four of these major layers are continuously “talking” to each other, modifying each other’s activity. Therefore, a slightly more detailed (and less misleading) way of analyzing the brain’s structure is by thinking of its architecture as falling into three broad functional levels, with multiple anatomical systems working in concert to produce specific effects.
– Autonomic control mechanisms of breathing, metabolism, homeostasis (medulla, reticular formation, and other elements of brainstem; hypothalamus; neuroendocrine system)
– Primary processing of raw sensory data (thalamus, localized regions of cortex)
– Proprioception (brainstem, sections of parietal lobe)
– Unconscious or pre-conscious emotional reactions (amygdala)
– Regulation of circadian rhythms (brainstem areas, especially reticular formation)
– Consolidation of raw sensory data into percepts or “objects” (localized regions of cortex)
– Planning movements (motor areas of cortex; basal ganglia)
– Coordination of motion in complex learned sequences (Cerebellum, motor cortex, basal ganglia)
– Attention (temporal lobe, fusiform cortex, frontal lobe)
– Desire and reward systems (ventral tegmental area; nucleus accumbens; prefrontal cortex)
– Emotions experienced as feelings (cingulate cortex; limbic system elements in temporal lobe)
– Pre-conscious “core” sense of selfhood as subjective “center” of experiences (brainstem, limbic system, various regions of cortex)
– Language processing (Broca’s area; Wernicke’s area; arcuate fasciculus)
– Processing of experiences into stored long-term memories
(hippocampus, various regions of cortex)
– Formation of concepts (regions of frontal lobe acting in concert with other areas of cortex and temporal lobe)
– Explicit self-awareness as a conscious agent (multiple systems acting in concert: anterior cingulate cortex, medial prefrontal cortex, insula, hippocampus, amygdala)
– Awareness of feelings (amygdala; orbitofrontal cortex; dorsal lateral prefrontal cortex; insula)
– Ability to retrieve memories at will (hippocampus, various regions of cortex)
– Ability to manipulate concepts in an ordered thought-process (multiple systems, particularly dorsolateral prefrontal cortex, orbitofrontal cortex, and elements of temporal lobe)
– Sense of situatedness in a continuum of past, present, and anticipated future (various regions, particularly frontal cortex and temperoparietal junction)
– Ability to make judgments about social appropriateness of behaviors (various regions of prefrontal cortex)
– Ability to learn new behavior patterns and to develop new modes of thinking (multiple systems, particularly dorsolateral prefrontal cortex, orbitofrontal cortex, and elements of temporal lobe)
II. What key aspects of the brain’s operations do scientists still not understand?
1. A new degree of complexity?
The computer scientist and psychologist John Holland, one of the founders of the field of genetic algorithms, gives an eloquent description of how much more complex the brain’s synaptic networks are than any other physical system ever studied.
It is unclear whether any, or all, of consciousness can be reduced to the interactions of neurons. … Indeed, we have neither theories, models, nor artifacts wherein each agent (neuron) simultaneously interacts with thousands of other agents (via synapses), and wherein the connections among agents involve so many feedback loops that a single agent may belong to hundreds or thousands of loops. Even an intricate computer provides only ten or so contacts for each component. To extrapolate from our current knowledge of machines to such machines is to jump three orders of magnitude in complexity. What we know of machines provides little guidance to machines of that complexity. … Until we know more about machines of this complexity, whether or not consciousness can be understood as an emergent property of certain kinds of machines is moot.
Holland’s humility here is instructive. He is not saying we cannot aspire someday to pick apart the brain’s higher-level operations and figure them out. He is merely observing that, given the current state of our scientific and engineering knowledge, that day is probably farther off than most people realize. The problem lies in the degree of complexity that the brain presents to us, as seekers of knowledge: we cannot be sure, but it may well present a conceptual challenge beyond anything the human mind has ever encountered in history.
Having said this, however, it is also worth observing the extraordinary, and accelerating, progress that fields such as neuroscience, cognitive psychology, molecular biology, and engineering have made over recent decades. These disciplines have been garnering huge (and growing) resources, and they have already made impressive breakthroughs in understanding physical systems of staggering complexity. The brain’s mysteries are being incrementally penetrated by a wide array of researchers in many fields, and it is not unreasonable to speculate that it is probably a matter of decades – not centuries – before some of our most profound questions will be answered.
2. How memories are stored and retrieved.
It is one thing to say that the laying down of memories is mediated by the neuronal mechanisms of long-term potentiation and synaptic plasticity, and that the hippocampus plays a key role, as it interacts with other major systems in the temporal and frontal lobes of the cortex. But scientists have yet to give a satisfactory account of precisely how a particular memory is encoded in a particular network of neurons, and how that pattern is accessed and retrieved on demand by other brain systems.
3. The binding problem.
Scientists are still not sure how, when I sit under a tree on a summer day and bite into an apple, my experience of this event comes together in a coherent whole. They know a lot about the mechanisms that characterize the various underlying components of this experience – how taste sensations are processed, how I situate myself in my spatial surroundings, which brain regions provide the requisite information for memory and context – but they have yet to understand the mechanisms through which all these disparate elements are bound together in a unified whole. How does the brain know which elements to include and which to exclude as it constructs a particular mental image, or experience?
4. Working memory, the mental space in which thoughts are held and manipulated.
Scientists know that certain regions of the frontal lobe play a key role here, but they have yet to offer a satisfying explanation for the mechanisms by which we consciously coordinate and execute our own thought process. They speak of “convergence zones” where synaptic signals from lower-order systems come together for processing, but they are still in the dark about how this convergence actually works.
Antonio Damasio, in his book The Feeling of What Happens, has offered the most satisfying hypothesis thus far about how multiple brain systems come together to construct an experiencing, conscious subject or self. He conceives of the process as a layered phenomenon, with basic or elemental forms of awareness providing the foundation, and an ascending hierarchy of mental states coming into play as one approaches full-scale self-conscious thought.  Here is the structure that Damasio envisions:
Autobiographical self and extended consciousness
Autobiographical memory and working memory
Second-order map of organism-object relationship
Images of object
Wakefulness, minimal attention
Damasio provides a clear definition of each element in this hierarchy of mental functions, and offers detailed correlations between each component in the hierarchy and a concomitant set of brain regions that are associated with generating it. In addition, he illustrates his description with clinical case studies of patients who have experienced various forms of brain damage, leading to the loss or alteration of particular elements in the ascending structure of mental states. Nevertheless, while his contribution marks an important milestone in understanding how a brain begets a mind, the key questions still remain: how, precisely, do all these systems coordinate their activities, moment-to-moment? What mechanisms govern the integration of these multifarious faculties into the seamless whole that each of us experiences as a given in our daily life?
Various neuroscientists – Damasio, Joseph LeDoux, Gerald Edelman, Rodolfo Llinás, Wolf Singer, Francis Crick and Christof Koch, to name a few – have proposed theories that seek to address these questions.  But those theories have yet to be borne out by detailed empirical research. Damasio summarizes the challenge that lies ahead:
Many of us in neuroscience are guided by one goal and one hope: to provide, eventually, a comprehensive explanation for how the sort of neural pattern that we can currently describe with the tools of neurobiology, from molecules to systems, ever becomes the multi-dimensional, space-and-time-integrated image we are experiencing this very moment. The day may come when we can explain satisfactorily all the steps that intervene from neural pattern to image but that day is not here yet.1
The cognitive scientist Douglas Hofstadter argues that our analysis of how a brain begets a mind will remain incomplete until we can account for the gradual integration, over years and decades, of elemental selfhood into full-fledged personhood. He argues that the higher faculties of a normally functioning mind should not be regarded as occurring in a mere snapshot of the brain’s activity at any one moment. Rather, they develop incrementally through a long process of developmental give-and-take, embedded in a social context.2 The mind/brain system, in other words, is a profoundly diachronic entity, and only emerges gradually through a laborious process of repeated self-adjustment, coupled with adjustment to external factors, over time. Hofstadter refers to this kind of entity as a ‘strange loop’: it only arises through years and years of iterative feedback-loops of trial-and-error self-modification. Another word for ‘strange loop,’ in this sense, is the much simpler term, ‘person.’ A normal human mind – all the faculties we normally associate with such a mind – cannot be divorced from the personhood that it simultaneously creates and that in turn re-creates and shapes it over a lifetime of experiences and actions.
 LeDoux, Synaptic Self; Bear, Neuroscience; Antonio Damasio, The Feeling of What Happens: Body and Emotion in the Making of Consciousness (Harcourt, 1999); Christof Koch, The Quest for Consciousness: A Neurobiological Approach (Roberts, 2004).
Damasio, The Feeling of What Happens; LeDoux, Synaptic Self; Gerald Edelman, Second Nature: Brain Science and Human Knowledge (Yale U. Press, 2006); Roberto Llinás and D. Paré, “Of dreaming and wakefulness,” Neuroscience 44 (1991): 521-533; Wolf Singer, et. al., “Formation of cortical cell assemblies,” Symposia on Quantitative Biology 55: 929-952; Koch, Quest for Consciousness.
Hofstadter, I Am a Strange Loop, especially chapter 18.