Module 7
Learning, Memory and Cognition
a. Learning as the Storage of Memories
Some one-celled animals “learn” surprisingly well, for example, to avoid
swimming toward a light where they have received an electric shock before. We have
placed the term learn in quotation marks because such simple organisms lack a
nervous system; their behavior changes briefly, but if you take a lunch break during
your subject’s training, when you return, you will have to start all over again. Such a
temporary form of learning may help an organism avoid an unsafe area long enough
for the danger to pass or linger in a place where food is more abundant. But without
the ability to make a permanent record, you could not learn a skill, and experience
would not help shape who you are. We will introduce the topic of learning by
examining the problem of storage.
H. M.’s symptoms are referred to as anterograde amnesia, an impairment in
forming new memories. (Anterograde means “moving forward.”) This was not H.
M.’s only memory deficit; he also experienced retrograde amnesia, the inability to
remember events prior to impairment. His retrograde amnesia extended from the time
of surgery back to about the age of 16; he had a few memories from that period, but
he did not remember the end of World War II or his own graduation, and when he
returned for his 35th high school reunion, he recognized none of his classmates.
Nevertheless, his memory was better for earlier events than for recent ones; this may
seem implausible, but it is typical of patients such as H. M. How far back the
retrograde amnesia extends depends on how extensively the medial temporal lobes are
damaged and whether the lateral temporal cortex is involved.
Not only did Henry Molaison devote much of his life to numerous scientific
investigations; his brain will continue to be the subject of study for many years to
come. Soon after his death, Henry’s preserved brain was in a plastic cooler strapped in
a seat on a flight from Boston to San Diego; in the next seat was Jacopo Annese,
director of the Brain Observatory at the University of California at San Diego. After
several months of preparation, Annese and his colleagues dissected Molaison’s brain
into slices as thin as the width of a hair (70 µm). The 53-hour, uninterrupted
procedure was recorded and live-streamed over the web to allow scientific scrutiny
and to increase public awareness and engagement (Annese et al., 2014). Each slice of
H. M.’s brain was microscopically photographed with such resolution that the data
from each one would fill 200 DVDs. The data were then combined into a
threedimensional reconstruction of the brain, which is available online. Scientists can
navigate through it to the area of their interest and then zoom in to the level of
individual neurons. H. M.’s memory problems made him perhaps the most studied
subject in neuroscience. Ironically, the man who could not remember will never be
forgotten.
The hippocampus consists of several substructures with different functions.
The part known as CA1 provides the primary output from the hippocampus to other
brain areas; damage in that part of both hippocampi results in moderate anterograde
amnesia and only minimal retrograde amnesia. If the damage includes the rest of the
hippocampus, anterograde amnesia is severe. Damage to the entire hippocampal
formation results in retrograde amnesia extending back 15 years or more. More
extensive retrograde impairment occurs with broader damage or deterioration, like
that seen in Alzheimer’s disease, Huntington’s disease, and Parkinson’s disease,
apparently because memory storage areas in the cortex are compromised.
H. M.’s memory impairment consisted of two problems: consolidation of new
memories and, to a lesser extent, retrieval of older memories. Consolidation is the
process in which the brain forms a more or less permanent physical representation of
a memory. Retrieval is the process of accessing stored memories—in other words, the
act of remembering. When a rat presses a lever to receive a food pellet, or a child is
bitten by a dog, the experience is held in memory at least for a brief time. But just like
the phone number that is forgotten when you don’t reach your party the first time you
dial, an experience does not necessarily become a permanent memory; if it does, the
transition takes time. Until the memory is consolidated, it is particularly fragile. New
memories may be disrupted just by engaging in another activity, and even older
memories are vulnerable to intense experiences such as emotional trauma or
electroconvulsive shock treatment (a means of inducing convulsions, usually in
treating depression). Researchers divide memory into two stages, short-term memory
and long-term memory. Long-term memory, at least for some kinds of learning.
The hippocampal area is not the permanent storage site for memories. If it
were, patients like H. M. would not remember anything that happened before their
damage occurred. According to most researchers, the hippocampus stores information
temporarily in the hippocampal formation; then, over time, a more permanent
memory is consolidated elsewhere in the brain. A study of mice that had learned a
spatial discrimination task supported the hypothesis: Over 25 days of retention testing,
metabolic activity progressively decreased in the hippocampus and increased in the
cortical areas. Although completing this transition takes time, the memory is
represented in prefrontal cortex (PFC) cells quickly. Fear conditioning in mice
activated specific cells in the hippocampus and the PFC. The next day, placing the
mice back in the chamber in which they had received shock activated those cells in
the hippocampus but not in the PFC. However, stimulating cells in either location
elicited the learned fear response (“freezing”), indicating that the PFC cells already
held a “silent” memory. Two weeks later, the PFC cells had developed increased
synaptic connections, and the hippocampal cells had lost connections; when the mice
were placed in the chamber, the PFC cells became active and mice froze, but the
hippocampal cells remained silent, indicating that responsibility for the memory had
been transferred to the PFC. This progression apparently takes longer in humans.
Christine Smith and Larry Squire (2009) used fMRI to image the brain’s activity
while human volunteers recalled news events from the past 30 years. Activity in the
hippocampus was greatest as they recalled recent events, declined as they recalled
events as far back as 12 years, and remained stable after that. Beyond the
hippocampus—in the prefrontal, temporal, and parietal cortex—activity increased
progressively with increasing age of the memory. So your brain works rather like your
computer when it transfers volatile memory from RAM to the hard drive—it just
takes longer.
Learning researchers were in for a revelation when they discovered that H. M.
could readily learn some kinds of tasks (Corkin, 1984). One was mirror drawing, in
which the individual uses a pencil to trace a path around a pattern, relying solely on a
view of the work surface in a mirror. H. M. improved in mirror-drawing ability over 3
days of training, and he learned to solve the Tower of Hanoi problem. But he could
not remember learning either task, and on each day of practice, he denied even having
seen the Tower puzzle before. What this means, researchers realized, is that there are
two categories of memory processing. Declarative memory involves learning that
results in memories of facts, people, and events that a person can verbalize or declare.
For example, you can remember being in class today, where you sat, who was there,
and what was discussed. Declarative memory includes a variety of subtypes, such as
episodic memory (events), semantic memory (facts), autobiographical memory
(information about oneself), and spatial memory (the location of the individual and of
objects in space). Nondeclarative memory involves memories for behaviors; these
memories result from procedural or skills learning, emotional learning, and
stimulusresponse conditioning. Learning mirror tracing or how to ride a bicycle or
solve the Tower of Hanoi problem are examples of nondeclarative learning or, more
specifically, procedural or skills learning; remembering practicing the tasks involves
declarative learning. Another way of putting it, which is admittedly a bit
oversimplified, is that declarative memory is informational, while nondeclarative
memory is more concerned with the control of behavior. Just as we have what and
where pathways in vision and audition, we have a what and a how in memory.
Conversely, rats with damage to the striatum could remember which arms they
had visited but could not learn to enter lighted arms. Because Parkinson’s disease and
Huntington’s disease damage the basal ganglia (which include the striatum), people
with these disorders have trouble learning procedural tasks, such as mirror tracing or
the Tower of Hanoi problem (Gabrieli, 1998). Incidentally, the term declarative seems
inappropriate with rats; researchers have often preferred the term relational memory,
which implies that the individual must learn relationships among cues, an idea that
applies equally well to humans and animals. You already know that the amygdala is
important in emotional behavior, but it also has a significant role in nondeclarative
emotional learning. Bechara and his colleagues (1995) studied a patient with damage
to both amygdalas and another with damage to both hippocampi. The researchers
attempted to condition an emotional response in the patients by sounding a loud boat
horn when a blue slide was presented but not when the slide was another color. The
patient with amygdala damage reacted emotionally to the loud noise, as indicated by
increased skin conductance responses. He could also tell the researchers which slide
was followed by the loud noise, but the blue slide never evoked a skin conductance
increase; in other words, conditioning was absent. The patient with hippocampal
damage showed an emotional response and conditioning, but he could not tell the
researchers which color the loud sound was paired with. This neural distinction
between declarative learning and nondeclarative emotional learning may well explain
how an emotional experience can have a longlasting effect on a person’s behavior
even when the person does not remember the experience.
The brain stores a tremendous amount of information, but information that is
merely stored is useless. It must be available, not just when it is being recalled into
awareness but when the brain needs it for carrying out a task. Working memory
provides a temporary “register” for information while it is being used. Working
memory holds a password you just looked up long enough for you to type it in; it also
holds information retrieved from long-term memory while it is integrated with other
information for use in problem solving and decision making. Without working
memory, we could not do long division, plan a chess move, or even carry on a
conversation.
b. Brain Changes in Learning
Almost a century ago, Donald Hebb (1940) stated what has become known as
the Hebb rule: If an axon of a presynaptic neuron is active while the postsynaptic
neuron is firing, the synapse between them will be strengthened. We saw this principle
in action during the development of the nervous system, when synaptic strengthening
helped determine which neurons would survive; some of that plasticity is retained in
the mature individual. Researchers have long believed that to understand learning as a
physiological process, they would have to figure out what happens at the level of the
neuron and particularly at the synapse.
Activity in presynaptic neurons also influences the sensitivity of nearby
synapses. If a weak synapse and a strong synapse on the same postsynaptic neuron are
active simultaneously, the weak synapse will be potentiated; this effect is called
associative long-term potentiation. Associative LTP is usually studied in isolated brain
tissue with artificially created weak and strong synapses, but it has important
behavioral implications, which is why it interests us. Electric shock evokes a strong
response in the lateral amygdala, where fear is registered, while an auditory stimulus
produces only a minimal response there. Rogan, Stäubli, and LeDoux (1997)
repeatedly paired a tone with shock to the feet of rats. Because of this procedure, the
tone alone began to evoke a significantly increased response in the amygdala, as well
as an emotional “freezing” response in the rats.
LTP has been studied most often in the neurons connecting CA1 and CA3 of
the hippocampus, and we will use those findings as our model here without going into
the variations that occur in other areas of the brain. LTP is induced through a cascade
of events at the synapse. In CA1 (and in most locations), the neurotransmitter
involved is glutamate, which is detected by two types of receptors: the AMPA (alpha-
amino-3- hydroxy-5-methyl-4-isoxazole propionic acid) receptor and the NMDA (N-
methyl-daspartate) receptor. Initially, glutamate activates AMPA receptors but not
NMDA receptors, because they are blocked by magnesium ions. During LTP
induction, activation of the AMPA receptors by the first few pulses of stimulation
partially depolarizes the membrane, and this dislodges the magnesium ions. The
resulting large influx of calcium ions activates a host of protein kinases, enzymes that
alter or activate other proteins. One of the protein kinases, CaMKII
(calcium/calmodulin-dependent kinase II), is required for LTP. Mice with two mutant,
nonfunctioning genes for the alpha form of CaMKII fail to show LTP; those with one
mutant and one functioning gene do show LTP, but it is not consolidated into long-
term memory.
Within 45 to 60 minutes after LTP, postsynaptic neurons develop dramatically
increased numbers of dendritic spines, outgrowths from the dendrites that partially
bridge the synaptic cleft and make the synapse more sensitive. Existing spines also
enlarge or split down the middle to form two spines. Another important structural
change is the appearance of new AMPA receptors, which increase synaptic strength.
These come from a pool of silent receptors that are transported into the spines from
within the dendrite; they can recycle between the cytoplasm and the membrane or in
the other direction within mere tens of minutes. A further change that occurs in
support of learning is the generation of new neurons in the hippocampus; though the
rate of neurogenesis is relatively low in adults, over the life span, new neurons add up
to an estimated 10% to 20% of the population. Numerous studies show that learning is
impaired by blocking neurogenesis and enhanced by increasing new cell birth. New
neurons are more active than mature ones, have a lower threshold for LTP induction,
and are better at making fine discriminations, such as distinguishing between the
contexts in which reward occurs and does not occur. It also appears that the new
neurons interfere with established LTP in the hippocampus, clearing the way for new
memories and aiding the transfer of memory to the cortex.
With all that growth, you might suspect that there would be some increase in
the volume of the brain areas that are involved in LTP. In fact, this does happen to
some extent. London taxi drivers, who are noted for their ability to navigate the city’s
complex streets entirely from memory, spend about 2 years learning the routes before
they can be licensed to operate a cab. Maguire and her colleagues (2000) used MRI to
scan the brains of 16 drivers. The posterior part of their hippocampi, known to be
involved in spatial navigation, was larger than in males of similar age. (Overall
hippocampal volume did not change; their anterior hippocampi were smaller.) The
difference was greater for cabbies who had been driving for the longest time, which
we would expect if the difference was due to experience.
The hippocampus can acquire learning “on the fly” while the event is in
progress, but a longer time is needed for long-term storage of declarative memories in
the cortex. Many researchers now believe that the hippocampus transfers information
to the cortex during times when the hippocampus is less occupied, even during sleep.
During sleep, neurons in rats’ hippocampus and cortical areas repeat the pattern of
firing that occurred during learning. Human EEG and PET studies showed the
hippocampus repeatedly activating the cortical areas that participated in the daytime
learning, and this reactivation was accompanied by significant task improvement the
next morning without further practice. Even a daytime nap is enough to evoke this
kind of activity in the hippocampus and improve subsequent performance. Anne
Schapiro and her colleagues (2018) collected fMRI data while their volunteers
reviewed the images they had just learned, then kept the scanner running as they
napped; comparing the data showed that images that had been remembered less well
during the post-training test were singled out for extra replay, which resulted in
improved recall later. Presumably, “offline” replay provides the cortex with the
opportunity to undergo LTP at the more leisurely pace it requires. During sleep, more
than 100 genes increase their activity, and many of those have been identified as
major players in protein synthesis, synaptic modification, and memory consolidation.
As hard as the brain works to make memories “permanent, ” it is still
important that these records not be inscribed in stone. Things change; the waterhole
we learned was reliable over several visits is now becoming progressively more
stagnant, so we must range in other directions until we find a new source of water.
And sometimes erroneous learning must be corrected; the first two redheads we knew
were hot tempered, and it will take meeting additional redheads to change what we
have learned. A memory needs to be stable to be useful, but at the same time it must
remain malleable; there are several ways the brain accomplishes this.
Most memories dissipate at least somewhat over time if they are not used
regularly. We usually think of forgetting as a defect, but it can also be useful. The
brain actively removes useless information to prevent the saturation of synapses with
information that is not called up regularly or has not made connections with other
stored memories. One way this active forgetting occurs involves an unexpected effect
of dopamine. During learning, some dopaminergic neurons facilitate learning while,
simultaneously, another subset causes the memory-modified neurons to release a
cascade of proteins that reduce the size and density of the postsynaptic dendrites. This
effect drops off shortly after learning occurs, which suggests that its role is to erase
memories that don’t receive further reinforcement. In addition, certain neurons in the
hypothalamus (melanin concentrating hormone neurons, MCH) inhibit activity in
hippocampal neurons. Stimulating these neurons had no effect on learning, but it
interfered with memory if administered later, suggesting a role in forgetting (Izawa et
al., 2019). MCH neurons are most active during rapid eye movement sleep, a stage of
sleep during which dreaming occurs and unneeded memories are removed.
These total recallers did no better than controls when they were asked to recall
personal events from a particular day during the past week, such as what they ate for
breakfast and what they did during the day. But their memories from the month before
were almost as good, and they could provide rich details from 1 year and 10 years
earlier, while controls remembered almost nothing. Apparently they’re not better
learners, but they forget less. Interestingly, total recallers scored as high on an
obsession inventory as patients with obsessive compulsive disorder (OCD). MRI
scans revealed enlargement of the caudate nucleus and putamen, both of which are
involved in OCD. Some of them are “germophobes,” while others collect things, such
as years of old TV Guides or hundreds of recorded TV shows; one arranges all the
bills in his wallet according to the city of the Federal Reserve Bank where they were
issued and how the sports teams in that city did. The researchers suggest this
compulsiveness also manifests itself in unintentional rehearsal of past events, which
strengthens the memories. Unfortunately, the researchers have not investigated the
alternative possibility that total recallers have inadequate active forgetting
capabilities.
c. Learning Deficiesncies and Disorders
Learning may be the most complex of human functions. Not surprisingly, it is
also one of the most frequently impaired. Learning can be compromised by accidents
and violence that damage the structures we have been studying. But more subtle
threats to learning ability come from aging and from disorders of the brain, including
Alzheimer’s disease and other dementias, which we will discuss in this section.
You may or may not find humor in this old joke, but declining memory is
hardly a laughing matter to the elderly. The older person might mislay car keys, forget
appointments, or leave a pot on the stove for hours. Working memory and the ability
to retrieve old memories and to make new memories may all be affected. Although we
usually associate aging with brain cell loss, significant deficits occur only in the
midbrain, basal forebrain (lower frontal lobes), and some prefrontal areas. Some parts
of the prefrontal cortex and hippocampus also decline in volume, likely due to a
decrease in synaptic density. These areas, of course, are critical for learning, memory,
and cognitive functioning (Mora, 2013). Deficits occur at the molecular level as well.
One study, for example, examined the brains of deceased individuals and found 17
genes in the dentate gyrus of the hippocampus that undergo changed levels of
expression with aging (Pavlopoulos et al., 2013). Downregulation of one of these
genes results in less abundant production of the protein RbAp48 in humans and mice.
This protein turns out to be important for memory: Young mice engineered to produce
reduced RbAp48 showed dysfunction in the dentate gyrus and performed like old
mice on memory tests. In another study, aged mice were significantly impaired on a
learning task after just 1 day without practice, but in old mice given a gene that
inhibits active forgetting, performance was still robust 4 weeks later (Genoux et al.,
2002). If we could find simple, safe ways to manipulate gene expression in humans,
we could reduce many of the burdens of aging.
Substantial loss of memory and other cognitive abilities (usually, but not
necessarily, in the elderly) is referred to as dementia. The most common cause of
dementia is Alzheimer’s disease, a disorder characterized by progressive brain
deterioration and impaired memory and other cognitive abilities. Alzheimer’s disease
was first described by the neuroanatomist and neurologist Alois Alzheimer in 1906,
after autopsying the brain of a 56-year-old patient with memory problems.
Alzheimer’s is primarily a disorder of the aged, although it can strike early in life. Of
the nearly 5 million people in the United States with Alzheimer’s, 4.7 million are over
the age of 65. The earliest and most severe symptom is usually impaired declarative
memory. Initially, the person is indistinguishable from a normally aging individual,
though the symptoms may start earlier; the person has trouble remembering events
from the day before, mislays items, forgets names, and must search for the right word
in a conversation. Later, the person repeats questions and tells the same story again
during a conversation. As time and the disease progress, the person eventually fails to
recognize acquaintances and even family members. Alzheimer’s disease is not just a
learning disorder but a disorder of the brain, so ultimately most behaviors suffer.
Language, visual-spatial functioning, and reasoning are particularly affected, and
there are often behavioral problems, such as aggressiveness and wandering away from
home. Alzheimer’s researcher Zaven Khachaturian (1997) eloquently described his
mother’s decline: “The disease quietly loots the brain, nerve cell by nerve cell, like a
burglar returning to the same house each night” (p. 21). Eventually, Alzheimer’s is
fatal, and it is the sixth leading cause of death in the United States (Heron, 2019). If
all the dementias were counted together, they would rank as the third leading cause of
death, behind heart disease and cancer. Even those numbers could be low, because
doctors often report the immediate cause of death, such as heart or kidney failure,
instead of dementia.
There are two notable characteristics of the Alzheimer’s brain, though they are
not unique to the disease. Plaques are clumps of amyloid beta (Aβ), a type of protein,
that cluster among axon terminals and interfere with neural transmission. The main
component is Aβ42, so called because it is 42 amino acids long; Aβ42 is particularly
“sticky, ” so it clumps easily to form the plaques. Even in asymptomatic elderly
individuals, the presence of plaques is associated with selfreported changes in
cognitive function and reduced test performance (Sperling et al., 2020). Aβ42—or,
more specifically, the ratio of Aβ42 to Aβ40—leads to the accumulation of the protein
tau to form neurofibrillary tangles inside neurons; tangles are associated with the
death of brain cells. Development of plaques and tangles progresses through the brain
in a predictable succession of stages, beginning with the medial cortex and
progressing to the limbic areas, particularly the hippocampus, and then to the
neocortex, the outer layers of cortex responsible for our highest functions (H. Braak &
Braak, 1991). This accounts for the pattern of changing symptoms as the disease
advances. We’ve also learned that this sequence parallels a progressive variation in
two significant factors in those areas: the relative level of expression of inflammatory
genes and the balance between proteins that promote or inhibit the aggregation.
The first clue to a gene location came from a comparison of Alzheimer’s with
Down syndrome (Lott, 1982). Down syndrome individuals also have plaques and
tangles, and they invariably develop Alzheimer’s disease if they live to the age of 50.
Because Down syndrome is caused by an extra chromosome 21, researchers zeroed in
on that chromosome; there they found mutations in the amyloid precursor protein
(APP) gene (Goate et al., 1991). When aged mice were genetically engineered with an
APP mutation that increased plaques, both LTP and spatial learning were impaired
(Chapman et al., 1999). Three additional genes that influence Alzheimer’s had been
confirmed by the end of the 1990s; all those genes affect amyloid production or its
deposit in the brain. The ε4 allele of the APOE gene is particularly interesting because
it contributes to so many Alzheimer’s cases. It increases risk by three- to eight-fold
and is associated with plaques and tangles, but how it contributes to pathology is not
well understood. Two studies indicate that carriers without dementia have lower
cerebral blood flow and that 2- to 25-month-old children with the allele have reduced
growth in temporal and parietal areas, which are affected in patients with Alzheimer’s.
The four genes in the table account for just a little more than half the cases of
Alzheimer’s disease, so there are likely many rare or small-effect genes as well as
environmental influences. Discovery of additional genes had to await whole-genome
studies with large numbers of individuals. Such studies have the advantage that they
allow gene searches without the need for a preconceived target area. Prior to 2009, 11
genes had been associated with Alzheimer’s, but a whole-genome study of 74,000
individuals was able to add 11 additional gene locations (Lambert et al., 2013).
Although the genes themselves have not been identified yet, genes near the loci are
involved in amyloid and tau pathways, inflammation, immune response, cell
migration, and cellular functions. Genome-wide studies have also made it possible to
do broad searches for epigenetic changes, and in the past few years, the focus has
been shifting in that direction.
A recent study conclusively identified seven genes that were differently
methylated in Alzheimer’s patients by taking the unusual step of verifying their results
in a second group of subjects (P. L. De Jager et al., 2014). Identifying the
environmental conditions that trigger these changes could help us reduce the
incidence of Alzheimer’s. A metaanalysis indicated several environmental risks for
dementia, including exposure to pesticides, fertilizers, herbicides, and insecticides;
airborne particulate matter; secondhand smoke; and electromagnetic fields. We’ve
also learned that neurons can modify their RNA through a process called somatic
recombination. We’ve known for years that immune cells use this strategy to adapt to
new threats. Ming-Hsiang Lee and colleagues (2018) discovered that neurons from
the brains of Alzheimer’s patients had six times as many varieties of the APP gene as
in healthy people, including 11 mutations already known to be linked to the inherited
form of Alzheimer’s.
Researchers are desperately seeking a cure for Alzheimer’s. Unfortunately, the
five drugs that are currently approved for the treatment of Alzheimer’s in the United
States offer benefits so small that they often don’t justify the side effects. Three of the
drugs are cholinesterase inhibitors; they restore acetylcholine transmission by
interfering with the enzyme that breaks down acetylcholine at the synapse.
Acetylcholine is important for attention, memory, and motivation, and
acetylcholinereleasing neurons are significant victims of degeneration in Alzheimer’s
disease. The fourth drug, memantine (marketed in the United States as Namenda), was
the first approved for use in patients with moderate and severe symptoms. Some of
the neuron loss in Alzheimer’s occurs when dying neurons trigger the release of the
excitatory transmitter glutamate; the excess glutamate produces excitotoxicity,
overstimulating NMDA receptors and killing neurons. Memantine limits the neuron’s
sensitivity to glutamate, reducing further cell death. Studies indicate moderate
slowing of deterioration and improvement in symptoms. Unfortunately, these drugs
treat only the symptoms, not the underlying cause, and they are little or no help when
degeneration is advanced.
preventing tau from forming tangles. After a string of failures with other anti-
amyloid drugs, most of the patients treated with aducanumab were declared amyloid
free; unfortunately, there was only minimal evidence of symptom improvement
(Sevigny et al., 2016). A later Phase 3 trial was halted for lack of results, but the
researchers then analyzed the data of patients who had chosen to continue taking the
drug and found the results were positive enough that the manufacturer is asking the
FDA to make aducanumab the first Alzheimer’s drug to receive approval since 2003
(P. Anderson, 2019). Two other research teams also received surprises after their
Phase 3 trials had failed to show improvements. They had made the unusual choice of
giving patients in their control groups tiny “placebo” doses of their tau inhibitors
(hydromethylthionine in one case and LMTM in the other). In both studies, patients
receiving the 25-times stronger therapeutic doses were unimproved at the end of the
trials, but the controls had dramatic reductions in brain atrophy and significant
improvement in cognitive functioning.
Further studies, of course, are underway to confirm the results and determine
the most effective dosages. A novel treatment was suggested by a study with mice
engineered to produce abundant plaques. Stimulating their brains with pulses of
ultrasound activated glial cells and reduced plaque levels, presumably because the glia
ingested them. As a result, the mice regained their lost memory capability. When
patients were treated with ultrasound, memory-related areas in their brains became
more active and their test performance improved for up to 3 months (Beisteiner et al.,
2020). Whether pathology progresses from amyloid plaques to tangles depends on
inflammation. The accumulation of Aβ triggers an immune response in microglia,
which release the inflammatory agent NLRP3 inflammasome (Ising et al., 2019). This
results in chemical changes in tau that cause it to become sticky and form into clumps.
We’ve known about the association between neural inflammation and Alzheimer’s for
years, but now NLRP3 provides a potentially important new target for reducing tau.
The aging individual dealing with memory problems typically wonders, “Am I
getting Alzheimer’s?” No single test can diagnose Alzheimer’s. Most often, the
individual will take a battery of memory and cognitive tests; the pattern of deficits
will help identify the cause as Alzheimer’s while ruling out other dementias. The
physician may also order an MRI to look for atrophy in the temporal and parietal
areas or a PET scan to identify areas of reduced activity. Using new tracers that target
Aβ and tau, PET scans can identify 75% to 90% of individuals who are later
confirmed by autopsy to have Alzheimer’s (Fraller, 2013). Biological assessment
should lead to more appropriate treatment planning by differentiating better between
Alzheimer’s and other dementias; in addition, it will be possible to monitor
therapeutic progress, detecting changes before they have time to translate into
cognitive gains. But because current treatments only slow the progress of
Alzheimer’s, researchers are interested in detecting the disease well before it is
fullblown and before irreversible damage has occurred. PET scanning for amyloid
predicts about one third of individuals with mild cognitive impairment who will be
diagnosed with Alzheimer’s during the next several months; among the normal
elderly, 25% with high amyloid levels are diagnosed within 3 years, while those with
low levels have a 98% chance of remaining cognitively stable (Gelosa & Brooks,
2012). Detection of Aβ and tau in cerebrospinal fluid has shown 98% to 100%
accuracy in identifying individuals who developed Alzheimer’s within the next 5 to 6
years. Although this test differentiates Alzheimer’s from fronto-temporal dementia, it
does not distinguish some others, such as Lewy body dementia.
Alzheimer’s accounts for 60% to 80% of all cases of dementia. Second is
vascular dementia, also called vascular cognitive impairment, at 5% to 10%. It is
caused by loss of blood supply to a part of the brain, for example, by stroke; in
memory-related areas, this produces Alzheimer’s-like symptoms, and in other areas,
the result can include confusion, motor problems, and language impairment. The third
most common type of dementia is fronto-temporal dementia. It is most often caused
by the accumulation of tau, or of the protein TDP-43. It differs from Alzheimer’s in its
greater language impairment and later onset of memory problems (Alzheimer’s
Association, n.d.-b). A less common (though likely underdiagnosed) form of dementia
is Korsakoff syndrome, brain deterioration that is almost always caused by chronic
alcoholism. The damage is due to a deficiency in thiamine (vitamin B1), which is
needed to produce energy from sugar. Alcohol reduces thiamine absorption in the
stomach, and the situation is made worse because the alcoholic tends to forgo food in
favor of alcohol. Korsakoff syndrome impairs declarative memory, especially
anterograde, without affecting nondeclarative forms. Thiamine therapy can relieve the
symptoms of Korsakoff syndrome somewhat if the disorder is not too advanced, but
the brain damage itself is irreversible. The hippocampus and temporal lobes are
unaffected in Korsakoff syndrome, but the mammillary bodies and the medial
thalamus are reduced in size, and structural and functional abnormalities occur in the
frontal lobes. A bizarre accident demonstrated that damage limited to the thalamic and
mammillary areas can cause anterograde amnesia; a 22-year-old college student
received a penetrating wound to the area when his roommate accidentally thrust a toy
fencing foil up his left nostril, producing an amnesia that primarily affected verbal
memory.
d. The Nature of Intelligence
Understanding how we measure intelligence is important because we are in
effect defining intelligence as what that test measures. The measure of intelligence is
typically expressed as the intelligence quotient (IQ). The term originated with the
scoring on early intelligence tests designed for use with children. The tests produced a
score in the form of a mental age, which was divided by the child’s chronological age
and multiplied by 100. The tests were designed to produce a score of 100 for a child
performing at the average for his or her chronological age. The scoring is completely
different now, partly because the tests were extended to adults, who do not increase
consistently in intellectual performance from year to year. The base score is still 100,
a value selected arbitrarily and preserved artificially by occasional adjustments to
compensate for any drift in performance in the population.
Claiming that true intelligence is much more than what the tests measure,
these critics often point to instances where practical intelligence or “street smarts” is
greater than conventional intelligence. For example, as young Brazilian street vendors
conducted their business, they were adept at performing calculations that they were
unable to perform in a classroom setting. In another study, expert racetrack gamblers
used a highly complex algorithm involving seven variables to predict racetrack odds,
but their performance was unrelated to their IQ; in fact, four of them had IQs in the
low to mid-80s. Robert Sternberg (2000) compared the scores that the presidential
candidates George W. Bush, Al Gore, and Bill Bradley made on the verbal section of
the Scholastic Aptitude Test (SAT) when they applied for college; the SAT has many
items similar to those on conventional intelligence tests. Two of the candidates scored
above average for college applicants but not markedly so, and one had a score that
was below average. To Sternberg, their success raises questions about the narrowness
of what intelligence tests measure. Sternberg (1988) argues that intelligence does not
exist in the sense we usually conceive of it but is “a cultural invention to account for
the fact that some people are able to succeed in their environment better than others”
(p. 71). Perhaps intelligence is, like the mind, just a convenient abstraction we
invented to describe a group of processes. If so, we should not expect to find
intelligence residing in a single brain location or even in a neatly defined network of
brain structures. And to the extent we find processes or structures that are directly
involved in intellectual ability, their activity might be only moderately correlated with
scores on traditional intelligence tests.
Another controversy that is critical to a biological understanding of
intelligence is whether intelligence is a single capability or a collection of several
independent abilities. Intelligence theorists tend to fall into one of two groups,
lumpers or splitters. Lumpers claim that intelligence is a single, unitary capability,
which is usually called the general factor, or simply g. General factor theorists admit
that there are separate abilities that vary somewhat in strength in an individual, but
they place much greater weight on the underlying g factor. They point out that a
person who is high in one cognitive skill is usually high in others, so they believe that
a measure of g is adequate by itself to describe a person’s intellectual ability. General
intelligence is often assessed by the overall IQ score from a traditional intelligence
test, such as the Wechsler Adult Intelligence Scale, whose 11 subtests measure more
specific abilities. But many g theorists prefer to use tests like the Raven Progressive
Matrices, because they emphasize reasoning and problem solving and are relatively
freer of influence from specific abilities such as verbal skills.
e. The Biological Origins of Intelligence
With this background, we are now ready to explore the origins of intelligence.
Based on our introduction, we will avoid two popular assumptions—that intelligence
tests are the only way to define intelligence and that intelligence is a single entity.
Instead, we will consider performance and achievement as additional indicators of
intelligence, and we will first examine the evidence of a biological basis for a general
factor and then consider the relationship between brain structures and individual
abilities.
Is a more intelligent brain different in any identifiable way from other brains?
Anyone asking this question would naturally wonder how Albert Einstein’s brain was
different from other people’s. Fortunately, the famous scientist’s brain was preserved,
and it has been made available from time to time to neuroscientists. In cursory
examinations, it turned out to be remarkably unremarkable. In fact, at 1,230 grams
(g), it was 100 g under the weight for the average male of his age. The number of
neurons did not differ from normal, and studies have disagreed about whether the
neurons were more densely packed or the cortex was thinner, perhaps because the
samples were taken from different locations. One study found a higher ratio of glial
cells to neurons in the left parietal lobe. The comparison brains averaged 12 years
younger than Einstein’s at the time of death, and we know that glial cells continue
proliferating throughout life (T. Hines, 1998), but the number of glial cells was not
elevated in Einstein’s right parietal lobe or in either frontal sample.
Popular wisdom is that smart people have bigger brains, and that is true—to an
extent, at least. A meta-analysis of 88 studies including more than 8,000 individuals
found a correlation between intelligence and brain size of. This means that variations
in brain size account for only about 6% (0.24 2 ) of people’s differences in
intelligence. Even this small amount of relationship breaks down if we try to apply it
across species; elephants have much larger brains, and not many people think
elephants are smarter than we are. In fact, it has been difficult to explain differences in
intelligence across species. Various proposals have come into vogue and then been
abandoned; the number of neurons in the brain, for example, again favors elephants,
and the ratio of brain weight to body weight indicates that the marmoset monkey
should be smarter than humans. After meticulously counting neurons in numerous
animal brains, Suzana Herculano-Houzel (2017) concluded that the best correlate of
cognitive ability across species is the absolute number of neurons in the cerebral
cortex. She suggests that the number of neurons in the prefrontal cortex, which is so
important in learning, language, and other higher-order behaviors, would probably be
an even better measure, but she hasn’t collected the data yet.
We said earlier that the idea of general intelligence easily accommodated
individual components, known as specific factors. Its original proponent, the English
psychologist Charles Spearman, came up with the theory after he used the very
powerful statistical method he invented, called factor analysis, to analyze a multitude
of tests of mental ability. The procedure involves giving a group of people several
tests that measure cognitive abilities. Then correlations are calculated among all
combinations of the tests to locate “clusters” of abilities that are more closely related
with each other than with the others. The strong correlation across most of the
measures suggested that mathematical skills, verbal skills, artistic ability, reasoning
skills, and so on were all influenced by a common factor, general intelligence
(Spearman, 1904). But factor analysis has also been useful in identifying clusters of
more specific abilities. Three capabilities have emerged frequently as major
components of intelligence: linguistic, logical-mathematical, and spatial.
f. Deficiencies and Disorders of Intelligence
Although intelligence and cognitive abilities do typically decline with age, the
amount of loss has been overestimated. One reason is that people are often tested on
rather meaningless tasks, such as memorizing lists of words; when the elderly are
tested on meaningful material, the decline is moderate (Kausler, 1985). Another
reason for the overestimation is that early studies were cross-sectional, comparing
people at one age with different people at another age; you have already seen that
more recent generations have an IQ test performance advantage over people from
previous generations. When the comparison is done longitudinally—by following the
same people through the aging process—the amount of loss diminishes (Schaie,
1994). Also, the loss depends on the type of intelligence measured. Crystallized
intelligence—skills and overlearned knowledge—remains stable or improves;
vocabulary and general knowledge, for example, improve through the seventh decade
of life. Fluid intelligence, which includes problem solving, reasoning, and the ability
to process and learn new information, declines beyond the third decade. A major
component of fluid intelligence is processing speed, which includes the speed of
performing cognitive activities and the speed of motor responses. Processing speed
begins to decline in the third decade (Harada et al., 2013). Schaie (1994) found that
statistically removing the effects of speed from test scores significantly reduced
elderly individuals’ performance losses. We saw earlier that working memory is
especially important to intellectual capability. In a study of higher-order information
processing, speed accounted for 80% of the variance in general intelligence, leading
the researchers to suggest that general intelligence is little more than this capability.
Some of the loss in performance is due to lack of opportunity to use skills. In
one study, aged individuals regained part of their lost ability through skills practice,
and many of them returned to their pre-decline levels; they still had some advantage
over controls 7 years later (Schaie, 1994). Elderly people also improved in memory
test scores when their self-esteem was bolstered by presenting them with words that
depict old age in positive terms, such as wise, learned, and insightful (B. Levy, 1996).
Loss in performance that has a physical basis may be reduced if not reversed. Diet
appears to be one factor; age-related decline can be reduced by consuming fish more
than once per week (B. Qin et al., 2014) and by having a cup of tea every day, even in
individuals with the ApoE allele (Feng et al., 2016). Both of these foods have
antiinflammatory and antioxidant effects. Vitamin supplements might also help. In
elderly individuals with mild cognitive impairment, a combination of B vitamins
reduced the rate of brain atrophy 50% and slowed cognitive decline, compared with
controls given a placebo. The B vitamins slow gray matter atrophy by reducing
homocysteine, a toxic compound that is elevated in people with a diet high in animal
proteins.
The most common genetic cause of ID is Down syndrome, with a prevalence
of 1 in every 700 births (S. E. Parker et al., 2010). Usually caused by the presence of
an extra 21st chromosome (trisomy 21), Down syndrome typically results in
individuals with IQs in the 40 to 55 range, although some are less impaired. Recall
that the amyloid precursor protein gene that is involved in early-onset Alzheimer’s
disease is located on chromosome 21 and that it was discovered because Down
syndrome individuals also develop amyloid plaques. Ninety-five percent of people
with Down syndrome have the entire extra chromosome; a smaller number have the
extra chromosome only in some of the body’s cells, and a few have only an end
portion of the chromosome, which is attached to another chromosome. Mouse models
of Down syndrome have been very useful in studying the disorder and attempting new
treatments. The most frequently used model is the Ts65Dn mouse, engineered with a
third copy of 55% of chromosome 21. Ts65Dn mice were treated prenatally and in the
early postnatal period with a peptide that supports neural survival and development;
this prevented developmental delay and reduced long-term memory impairment in
adulthood. However, it may be an oversimplification to focus only on the effects of
chromosome 21 genes. DNA methylation differs by more than 10% between
individuals with Down syndrome and controls, and the degree of methylation
difference is correlated with cognitive function.
Autism Spectrum Disorder Autism spectrum disorder (ASD) is a set of
neurodevelopmental disorders characterized by social deficits, communication
difficulties, and repetitive behaviors. In DSM-5, ASD includes the previously used
diagnoses of autism, Asperger syndrome, pervasive developmental disorder not
otherwise specified, and childhood disintegrative disorder (American Psychiatric
Association, 2013). Although the term autism refers to a subset of these disorders, it is
used interchangeably with autism spectrum disorder in the profession; we will follow
suit in the following discussion, especially when describing older studies, which used
narrower diagnostic criteria. The prevalence of ASD was 1 in 150 children (.67%) in
the years 2000 and 2002 but increased to 1 in 68 by 2010 (1.5%) and appeared to
level off before rising to 1 in 54 (1.85%) in 2016, the most recent year for which we
have data. Rates in North America, Asia, and Europe are similar, averaging between
1% and 2%. At least some of the increase can be attributed to improved detection,
broader diagnostic criteria, and doctors’ greater willingness to use the label because of
decreasing stigmatization of autism and because the diagnosis will qualify the family
for increased services and financial assistance (Rutter, 2005). Some observers believe
the increase is real and due, for example, to rising environmental toxins, but the
position of most authorities is that we simply don’t know how much the actual rate
has increased or why. One statistic is not in dispute: The prevalence is more than four
times higher among boys than girls.
One third of children with ASD have ID, defined as IQ < 70; this percentage is
higher in girls than boys (40% versus 32%) and among black and Hispanic compared
to white children (47%, 36%, and 27%, respectively; Centers for Disease Control and
Prevention, 2020b). Individuals with autism share a common core of impairment in
communication, imagination, and socialization (Frith, 1993). Trouble understanding
verbal and nonverbal communication often makes testing difficult, raising questions
about the meaningfulness of test results. In some cases, nonverbal tests such as the
Raven Progressive Matrices are used, or IQ is estimated from an assessment of
adaptive behavior, but these do not eliminate the deficits. Difficulty with imagination
also is common, in the form of an inability to pretend or to understand make-believe
situations. Use of language is also very literal—“Can you pass me the salt?” is met
with “Yes” with no compliance—and in some cases, this literalism extends to an
obsessive interest in facts, like that seen in the movie Rain Man.
Some researchers believe that much of the social behavior problem is that the
person with autism lacks a theory of mind, the ability to attribute mental states to
oneself and to others. In other words, the person with autism cannot infer what other
people are thinking. One man with autism said that people seem to have a special
sense that allows them to read other people’s thoughts (Rutter, 1983), and an
observant youth asked, “People talk to each other with their eyes. What is it that they
are saying?”. In a study that measured this deficiency, children watched hand puppet
Anne remove a marble from a basket where puppet Sally had placed it and put it in a
box while Sally was out of the room. On Sally’s return, children were asked where
she would look for the marble. Normal 4-year-olds had no problem with this task, nor
did Down syndrome children with a mental age of 5 or 6. But 80% of children with
autism with an average mental age of 9 answered that Sally would look in the box.
There are two hypotheses as to how we develop a theory of mind. According to the
“theory theory, ” we build hypotheses over time based on our experience. Simulation
theory holds that we gain insight into people’s thoughts and intentions by mentally
mimicking the behavior of others.
Individuals who score higher on a measure of empathy tend to have more
activity in these mirror neurons. Researchers have suggested that impaired mirror
functions reduce the ability of a person with autism to empathize and to learn
language through imitation. Children with ASD engage in less contagious yawning
than other children do (Senju et al., 2007), and they show neural deficiencies during
mirroring tasks.. Other studies show reduced activation in the inferior frontal cortex
and motor cortex, suggesting weakness in the dorsal stream connections that provide
input to those areas. This interpretation was supported in a study of individuals
diagnosed with Asperger syndrome, the former term for high-functioning autism.
When they imitated facial expressions, transmission over the dorsal stream (occipital
to superior temporal to posterior parietal to frontal) was delayed by 45 to 60
milliseconds compared with normal controls (Nishitani et al., 2004).
excitable brain; the behavioral manifestations include inattention, excitability,
and seizures. There is evidence this overactivity may derive from an
excitation/inhibition imbalance due to excessive glutamate activity or a deficit of
GABA (gammaaminobutyric acid) activity; you may remember these as the brain’s
primary excitatory and inhibitory transmitters, respectively. There is little evidence
that GABA modulators treat the core symptoms, but therapeutic results have been
reported for drugs that alter functioning of the NMDA glutamate receptor. NMDA
receptor antagonists (memantine, amantadine) reduce social and cognitive deficits,
inattention, and hyperactivity and other symptoms. On the other hand, an NMDA
receptor agonist (D-cycloserine) reduces social withdrawal and repetitive behavior.
Apparently, deviations from normal functioning of the glutamate system in either
direction can produce different symptoms of ASD in different individuals. Serotonin
contributes to brain development, including synapse formation and pruning. About
one third of people with ASD have elevated blood levels of serotonin, but serotonin
synthesis is reduced in a subset of children with autism. Several studies attempted
treatment with fenfluramine, an indirect serotonin antagonist, but without success in
reducing core symptoms. Treatment with SSRIs, antidepressants that increase
serotonin availability at the synapses by blocking reuptake into the terminals, has
been ineffective with children and minimally effective with adults. These failures may
have occurred because the subject pools included individuals from both subgroups.
According to the dopamine hypothesis of ASD, deficits in social behavior are
due to reduced dopamine activity in the mesocorticolimbic reward system, and
stereotyped (repetitive, purposeless) motor behaviors are caused by excess dopamine
activity in the nigrostriatal circuit (substantia nigra to dorsal striatum; Pavăl, 2017).
Because both of these behaviors are also seen in schizophrenia and other psychoses,
doctors have tried using antipsychotic drugs to treat ASD. Two antipsychotics,
aripiprazole (brand name Abilify) and risperidone (Risperdal), are the only drugs
approved by the Food and Drug Administration to treat symptoms of ASD; they are
dopamine antagonists, which work by blocking certain dopamine receptors. They are
effective in reducing social withdrawal, repetitive behaviors, and hyperactivity. These
drugs are suitable for short-term use only, because their side effects include significant
weight gain and risk for producing involuntary movements, or tardive dyskinesia.
Parental treatment has been ruled out as the cause of autism, but numerous
other environmental conditions have been identified as contributing factors. One’s
suspicions immediately turn to environmental pollutants, such as those generated by
automobile traffic, agricultural practices, and industrial activity. Researchers usually
assess such effects by studying people living near agricultural fields or a busy
highway, but two studies are notable for employing more meaningful measures of
pollution. Using localized estimates of traffic pollution based on EPA data, traffic
volume, and climatic conditions during each child’s gestational period and first year
of life, researchers found that living in homes where traffic pollution was highest was
associated with triple the rate of autism. Recognizing that ASD occurs in geographic
clusters, University of Chicago researchers used county-level data on genital
malformation in male children as a surrogate for environmental exposure and found
that the incidence of ASD increased 283% for every percentage point increase in
incidence of malformations (Rzhetsky et al., 2014). There are numerous other
environmental risk factors as well. One is paternal age, possibly due to chromosomal
abnormalities that develop with aging. Risk was almost doubled in children of fathers
between the ages of 34 and 39 and more than doubled beyond the age of 40. Birth
complications can impair later brain development and lead to ASD; cesarean delivery
and umbilical cord around the neck increase the risk of autism by 26%. The mother’s
use of common medications during pregnancy, such as antidepressants or
acetaminophen, also raises the risk, as do her health issues, including hypertension
and diabetes. Particularly dangerous is maternal infection during the second trimester
of pregnancy, which triples the chance of a woman having a child with ASD (Zerbo et
al., 2015). The news is not all bad: Mothers who took folic acid during pregnancy
were half as likely to bear a child who would later be diagnosed with autism (Surén et
al., 2013). And the strongly held belief among some parents that ASD is linked to
childhood vaccines doesn’t hold up to scientific scrutiny, as A Further Look shows.
The list of environmental influences is so long it makes one wonder how much
is left for genetic influence. It turns out to be a lot. Siblings of children with autism
are 25 times more likely to be diagnosed with autism than other children (Abrahams
& Geschwind, 2008); the number would be even higher, but parents tend to stop
having children after the first autism diagnosis. For the identical twin of a child with
autism, the risk of autism is at least 60%; when autistic-like symptoms are considered
in the second twin, the concordance jumps to 92%, compared with 10% for fraternal
pairs (A. Bailey et al., 1995). In a study of the 7 million children born in Sweden
between 1982 and 2006, heritability was estimated at 83%. The variance that could be
attributed to nonshared environmental influence was 17% and for shared environment
was 4% (Sandin et al., 2017). A study of 2 million children from five countries
agreed, placing the heritability estimate at 81%. Interestingly, the values varied by
country, ranging from 51% to 87% (Bai et al., 2019). When we look at epigenetic
effects and de novo mutations that are not shared with parents and siblings, we realize
that genes have additional effects that are not recognized in heritability estimates. It
soon became clear that the genes responsible for ASD were numerous and had very
small effects; to make finding them all the more difficult, relatively few afflicted
people in a study will share the same genes. Genes that have been identified include
those responsible for proteins involved in synapse formation and the neurotransmitters
we identified earlier (along with others) and their receptors. The transmitters are
important not only for their role in neural communication but also because they
contribute to neuronal migration, differentiation, and synapse formation and pruning
during development.
Researchers have looked to genes for an explanation of the greater male risk
for ASD. The search for the obvious candidates, X-linked genes, has not been very
fruitful. There is, however, some genetic support for the “female protective model”
we mentioned earlier (Jacquemont et al., 2014). Females may gain a benefit from the
RORA protein, which regulates the activity of more than 2,500 genes; some of these
are involved in neuron and synapse development and synaptic transmission, and at
least six have been associated with ASD. The RORA gene is upregulated by estrogens
and downregulated by testosterone, and the protein is higher in females than in males.
Irene Voineagu and her colleagues (2011) identified more than 500 genes that were
expressed at different levels between the frontal and temporal cortices of healthy
brains and only eight in ASD brains. As you have seen before, environmental
influences almost certainly exert their effect through epigenetic modification of gene
expression; this fact offers a glimmer of hope in that drugs designed to reverse those
modifications might be used as a treatment for autism.
The source of the autistic savant’s enhanced ability is unknown. Dehaene
(1997) suggests that it is due to intensely concentrated practice, but more typically the
skill appears without either practice or instruction, as in the case of a 3-year-old girl
with autism who began drawing animated and well-proportioned horses in perfect
perspective (Selfe, 1977). Allan Snyder and John Mitchell (1999) believe that these
capabilities are within us all and are released when brain centers that control
executive or integrative functions are compromised. This, they say, gives the savant
access to speedy lower levels of processing that are unavailable to us. But lacking the
executive functions, the savants perform poorly on apparently similar tasks that
require higherorder processing. The idea gains some credibility from the case of a
man impaired in his left temporal and frontal areas by dementia; despite limited
musical training, he began composing classical music, some of which was performed
publicly.
If these savants have an island of exceptional ability, autism is an island of
impairment in high-functioning individuals with ASD. As an infant, Temple Grandin
would stiffen and attempt to claw her way out of her affectionate mother’s arms (O.
Sacks, 1995). She was slow to develop language and social skills, and she would
spend hours just dribbling sand through her fingers. A speech therapist unlocked her
language capability, starting a slow emergence toward a normal life. Even so, she did
not develop decent language skills until the age of 6 and did not engage in pretend
play until she was 8. As an adult, Grandin earned a doctorate in animal science; she
teaches at Colorado State University and designs humane facilities for cattle, while
lecturing all over the world on her area of expertise and on autism. Still, her theory of
mind is poorly developed, and she must consciously review what she has learned to
decide what others would do in a social situation. She says that she is baffled by
relationships that are not centered on her work and that she feels like “an
anthropologist on Mars.”
Nevertheless, concerns about treating young children with drugs has led to the
development of the first medical device to be approved for treating ADHD in
children. Used while sleeping, a pocket-sized stimulator delivers pulses of electricity
through wires to a small adhesive patch on the child’s forehead to stimulate the
trigeminal nerve, the cranial nerve that carries sensory information from the face to
the brain (J. J. McGough et al., 2019). Research indicated that the stimulation inhibits
overactive neurons and stimulates blood flow in areas that control mood, attention,
and executive function. Although ADHD is considered a childhood disorder, a 13-
year-long study in the United Kingdom found that 22% of the children with ADHD
continued to meet criteria at age 18 (Agnew-Blais et al., 2016). DSM-5 diagnostic
criteria require onset during childhood, but 68% of the 18-year-olds who met
diagnostic criteria without regard to age of onset had not done so in any of the four
previous screenings. These results are not unique to the United Kingdom; they were
confirmed in a study in Brazil and another in New Zealand (Caye et al., 2016).
Unfortunately, the data can’t tell us whether adultonset ADHD is a different form;
because the adult-onset group had higher IQs and were predominantly female, the
researchers suggested that the individuals might have been able to mask their
symptoms during childhood.
Numerous studies have reported one difference or another in the brains of
individuals with ADHD, but the results have been inconsistent. Brain MRIs cost
anywhere between $1,000 and $5,000, so study samples are usually small; brain
differences in psychological disorders are typically slight, on the order of just a few
cubic centimeters of volume, so reliable detection of these differences requires large
samples. Now a nine-country international group has solved that problem by
compiling data from 1,700 people diagnosed with ADHD and 1,500 people without
(Hoogman et al., 2017). They found reductions in overall brain volume and in five
specific structures involved in emotion, motivation, and reward processing: the
caudate nucleus, putamen, nucleus accumbens, amygdala, and hippocampus. The
amygdala had the most pronounced deficit, which is important because the amygdala
has functions not only in emotion but also in response inhibition. Other research has
found deficiencies persisting at 33-year follow-up.
To say that ADHD runs in families would be an understatement: Heritability
averages 76% across studies (Biederman & Faraone, 2005; Brikell, Kuja-Halkola, &
Larsson, 2015), and concordances have been variously estimated at 58% to 83% for
identical twins and 31% to 47% for fraternal twins (Wender, Wolf, & Wasserstein,
2001). Like autism, ADHD is a complex, multi-symptom disorder, and different
individuals display different combinations of symptoms. Whole-genome studies have
implicated genes involved in cell migration, synaptic excitability, and neuronal
plasticity (Jain et al., 2012). Another study found 222 copy number variations (CNVs)
in ADHD patients that were not found in control subjects (Elia et al., 2010). These
CNVs were in or near genes involved in synaptic transmission, neural development,
and learning and other psychological functions. Numerous CNVs are in chromosomal
locations that have also been identified in autism and schizophrenia (N. M. Williams
et al., 2010); participation of a gene in more than one disorder should not be
surprising, because different disorders share some symptoms. And remember that we
are talking not about a gene for autism or a gene for ADHD or a gene for
schizophrenia but about genes that regulate brain growth, receptor development, and
so on.
However, the genes most frequently implicated in ADHD are involved with
neurotransmission. In fact, it is difficult to discuss the genetics of ADHD without
simultaneously discussing neurotransmitters. A review of genetic studies identified 24
“hot genes,” each of which had been linked to ADHD in at least five studies. The
large majority of these genes are involved with dopamine, serotonin, and
norepinephrine transmission, including synthesis, transport, and receptors. In addition,
one gene’s protein contributes to the construction of the nicotinic acetylcholine
receptor. The most frequently prescribed drugs, methylphenidate and amphetamines,
increase dopamine and norepinephrine activity by blocking reuptake at the synapse.
This implicates both transmitters, but most research links ADHD to reduced activity
in dopamine pathways, especially in the prefrontal cortex and the striatum. These
structures’ functions include executive control, impulse inhibition, working memory,
movement, learning, and reward —all functions that are affected in ADHD. The
significance of reward may be less immediately obvious, but several researchers
believe that impaired reward contributes to impulsiveness because the allure of later
rewards is too weak to overcome the temptation of immediate gratification. Some
20% to 30% of patients do not respond to the traditional medications or cannot
tolerate them (Biederman & Faraone, 2005). Two drugs sometimes used in their place,
modafinil and atomoxetine, block norepinephrine reuptake, adding support for its role
in ADHD as suggested by genetic studies.
Known environmental contributors to the risk for ADHD include brain injury,
stroke, and complications during pregnancy and birth (Biederman & Faraone, 2005;
Castellanos & Tannock, 2002). Toxins such as lead are also a factor; although
eliminating lead in gasoline and paint has reduced levels in the environment, it is
occasionally found in children’s costume jewelry and imported candies, as well as in
the soil and water. A recent study confirmed higher blood levels of lead in children
diagnosed with ADHD, and lead levels were correlated with teacher and parent
ratings of symptoms. Since the contamination of the Flint, Michigan, water supply
with lead, the number of schoolchildren qualifying for special education services for
ADHD, dyslexia, and mild intellectual impairment has increased from 15% to 28%
(Green, 2019). As with autism, another culprit is organophosphate pesticides; children
with aboveaverage urinary levels of a metabolite of these pesticides are twice as likely
to be diagnosed with ADHD, compared with children with undetectable levels. One
organophosphate may be especially potent; the risk of ADHD increased—by as much
as three times—with increasing urinary levels of dimethylphosphate (C. J. Yu et al.,
2016). Other pesticides are also suspect; for example, pyrethroid levels in mothers at
gestational week 28 were associated with ADHD traits in their 2- to-4-year-old
children.
The maternal environment is also implicated. ADHD is correlated, for
example, with maternal smoking and stress during pregnancy. Even the mother’s use
of cosmetics, perfumes, and shampoo containing phthalates is turning out to be a
problem; mothers who had higher levels of phthalate metabolites in their urine during
the second trimester of pregnancy were more likely to have a child diagnosed with
ADHD; the rate tripled in mothers in the top 20% of phthalate exposure (S. M. Engel
et al., 2018). There is some preliminary evidence that phthalates decrease
neurogenesis and proliferation. It is important to remember that environmental insults
often exert their influence by changing gene functioning. For example, inducing
concussion in rats altered more than 1,200 genes, 16 of which are linked to ADHD (Q.
Meng et al., 2017). Another example is dietary: High fat and sugar in women’s diets
during pregnancy was associated with higher symptoms of ADHD and with increased
methylation of the IGF2 gene, which is involved in development of brain areas
implicated in ADHD (Rijlaarsdam et al., 2017). In addition, genes can influence
whether an environmental condition has an effect; in children with ADHD who had a
specific mutation in the C282Y gene, “safe” levels of lead exposure were associated
with increased symptoms.
