Content uploaded by Pietro Michelucci
Author content
All content in this area was uploaded by Pietro Michelucci on Mar 09, 2015
Content may be subject to copyright.
The Psychopathology of Information Processing
Systems
Matthew Blumberg1 & Pietro Michelucci2
1GridRepublic
2Human Computation Institute
Abstract Information processing systems composed of groups of humans may
exhibit modes of dysfunction that correspond to psychopathology observed in in-
dividuals. Thus, clinical models normally applied to individuals are considered as
candidate models for understanding psychosis and neurosis in distributed systems.
In the first part, Matthew Blumberg considers dysfunction at the interaction level
in the context of schizophrenia, and in the second part, Pietro Michelucci exam-
ines dysfunction at the neurological level in the context of obsessive-compulsive
disorder.
Introduction
If we organize systems of human participants modeled loosely on cognitive ar-
chitecture (see Blumberg, Heylighen, Pavlic and Pratt, all this volume), it is con-
ceivable that such systems will exhibit dysfunction, just as humans do. And it
therefore will be necessary to develop methods of thinking about, diagnosing, and
treating (e.g. debugging) these issues.
Two approaches will be explored: the first views mental illness from the stand-
point of communications theory and logical structure– essentially viewing mental
illness as failure in information processing. The second views mental dysfunction
from the standpoint of brain chemistry. Each maps somewhat differently to in-
formation processing systems— suggesting different modes of analysis and means
of intervention.
The goal of this chapter is to discuss the nature of such systemic pathologies
speculatively—we aim not to provide a rigorous analysis, but rather to begin a
conversation.
Note: This is an author pre-submission version of a chapter from P. Michelucci
(ed.), Handbook of Human Computation, DOI 10.1007/978-1-4614-8806-4, ©
Springer Science+Business Media New York 2013, published version available
at: http://link.springer.com/chapter/10.1007/978-1-4614-8806-4_6
2
Part I: Interaction Dysfunction
Schizophrenia in Persons
What is Schizophrenia?
Gregory Bateson describes a Schizophrenic as “a person who does not know
what kind of message a message is.” (Bateson, et al., 1956). In broad terms, this
means understanding the condition as being rooted in a failure to discern context;
more rigorously it means understanding the condition as a specific patterned fail-
ure to keep straight the “logical type” of messages. From this perspective, what
manifests as mental illness is at root a pathology of information processing.
The “Theory of Logical Types” (Whitehead and Russell, 1927) is a formal way
to describe what one might intuitively describe as “levels of abstraction” within a
set of information. In this formalization, each item of a set is a member of a
“class”. A critical distinction is that a class cannot be a member of itself; nor can
one of the members of the class be the class itself. That is, a “class” represents a
higher level of abstraction—a higher logical type—than its “members”. In other
words: information is hierarchical.
A few examples illustrate this idea of “logical types” (Bateson, 1979):
• The name is not the thing named but is of different logical type, higher
than the thing named.
• The injunctions issued by, or control emanating from, the bias of a house
thermostat is of higher logical type than the control issued by the ther-
mometer. (The bias is the device on the wall that can be set to determine
the temperature around which the temperature of the house will vary.)
• The word tumbleweed is of the same logical type as bush or tree: It is not
the name of a species or genus of plants; rather, it is the name of a class
of plants whose members share a particular style of growth and dissemi-
nation.
• Acceleration is of a higher logical type than velocity.
The use of Logical Typing can be seen to be fundamental to human communi-
cation. “Play”, “non-play”, “fantasy”, “sacrament”, “humor”, “irony”, and “learn-
ing” are all examples of classes of communication. In all these cases, proper inter-
3
pretation of a specific message1 depends upon proper identification of the Logical
Type to which it belongs – i.e., proper identification of the context of the message.
By way of example, children may “Play” at fighting. While such activity may
have many of the outward markers of “real” fighting, the participants can never-
theless engage without anger or malice. But if a participant for some reason
comes to see the context of such interaction as “real” fighting, the meaning of
events changes, and the actions may degrade to harmful violence.
Returning to our thesis, Bateson defines a schizophrenic as one who (a) has dif-
ficulty in assigning the correct communicational mode to the messages he receives
from other persons; (b) has difficulty in assigning the correct communicational
mode to those messages which he himself utters or emits nonverbally; and/or (c)
has difficulty in assigning the correct communicational mode to his own thoughts,
sensations, and percepts. Most generally – “He has special difficulty in handling
signals of that class whose members assign Logical Types to other signals.”
In “Toward A Theory Of Schizophrenia”, Bateson looks specifically at the role
“Double Binds”, situations in which messages implicit at different levels of com-
munication conflict with one another. These experiences can be very difficult for
affected persons to define, because conflict exists between different levels of the
same interaction. For instance,
A young man who had fairly well recovered from an acute schizophrenic
episode was visited in the hospital by his mother. He was glad to see her and
impulsively put his arm around her shoulders, whereupon she stiffened. He
withdrew his arm and she asked, ‘Don’t you love me any more?’ He then
blushed, and she said, ‘Dear you must not be so easily embarrassed and afraid
of your feelings.’ (Bateson, et al., 1956)
I.e., in this example the young man can become confused by the conflict be-
tween meanings implicit at different levels of the same interaction: the mother on
the one hand stiffening when hugged, while at same time asking “don’t you love
me anymore?”; and this is made worse when the mother provides the young man
with an incorrect attribution for his inner confusion (“Dear, you must not be so
easily embarrassed by your feelings.”)
The means by which such apparently small communication issues can lead to
large scale pathology is in some sense analogous to how the flow if a river creates
a canyon: not by brute force, but by the slow and persistent effect of the water’s
friction over time. Thus in the above example, imagine that young man, growing
up, has been subject to millions of similarly muddled communications over his en-
tire life– always in the vital emotional context of a parent-child relationship. The
young man, seeking to make sense of the world, may thus learn to muddle the log-
1 Note that “message” here is used in a general way: including verbal utterance, non-
verbal action, absence of utterance or action, etc. Anything that can or should be taken as
being of consequence in a social interaction.
4
ical type of messages-- as a means to adapt, to make sense of things. (It is, in a
sense, a perfectly rational response to an irrational environment). And the child,
having so learned, will take these interpretative habits into secondary relationships
as well (including his relationship with himself).
Schizophrenia in Social Groups:
In the above view, certain patterned failures in communication and interpreta-
tion lead to behavioral dysfunction. We wish to introduce, speculatively, the no-
tion that similar patterns of communication within social groups will lead to simi-
lar dysfunction -- at the group level. I.e., that there exist pathologies of
information systems generally, which can be exhibited at multiple scales, individ-
ual or group.
From the point of view described earlier, {1} Schizophrenia in an individual is
understood to be rooted in the individual’s inability “ to know what kind of mes-
sage a message is”; and {2} that this confusion is often related to repeated experi-
ence of double-binds. With this in mind, consider:
A characteristic feature of contemporary media has been the blending of news
and entertainment; of advertising and news; of opinion and fact, of expert and am-
ateur. That is: one who turns to media for information about the world literally
does not know what kind of messages he is getting.
Similarly, American political discourse is frequently characterized by double-
binds, often with each of the two primary political parties articulating conflicting
messages about the nature of reality. For instance, a terrorist attack creates a bind,
putting into conflict essential principles civil liberties with the desire for security.
The increasing unreliability of information type (is it news? entertainment? ad-
vertising?), frequently experienced with particular acuteness around issues pre-
sented by institutional parties as double-binds (ex the requirements of civil liber-
ties vs security), can thus be recognized as a communications pattern very much
like that experienced by the individual schizophrenic. And so it is perhaps not
surprising that the adventurous cultural analyst might perceive comparable symp-
toms to be exhibited at the cultural level. I.e., one might consider contemporary
culture to be exhibiting certain “schizophrenic” patterns— for instance: pervasive
belief in conspiracy theories2 (i.e. paranoia); a government with reduced capacity
to take consistent decisive action; economic dysfunction.
2 “About half the American public endorses at least one kind of conspiratorial
narrative” -- "Conspiracy Theories, Magical Thinking, and the Paranoid Style(s)
of Mass Opinion". J. Eric Oliver and Thomas J. Wood, Working Paper Series,
University of Chicago, 2012
5
As problems from this point of view are understood to be rooted in deficiencies in
communications, interventions would be similarly focused. Examples might
include actions to strengthen the journalistic establishment; to alter forms of
political discourse, especially to either minimize or explicitly recognize the
double-binds; and perhaps to introduce “therapeutic” double-binds, the resolution
of which would require denial of one or another element of a larger bind.
Part II: Organismal Dysfunction
Next we apply an organismic view to complex, distributed information-
processing systems, endowing them with agency, such as goal-directed behavior,
and a tendency toward homeostasis – an equilibrium state. This view permits us
to adapt extant pharmacological treatment models of human behavioral dysfunc-
tion to neurosis in these distributed systems.
An Associative View
In general, thought-processes in humans are influenced heavily by fundamental
drives. According to drive theory (Seward, 1956), any disturbance to the equilib-
rium state in a person drives the person to engage in thought processes leading to
behaviors that restore homeostasis, to ensure survival. For example, dehydration
gives rise to thirst, which drives a person to seek water (the goal state). This leads
to a series of thoughts about how to obtain water that might involve planning and
decision-making (see Busemeyer and Townsend, 1993). Though drive theory is
oft criticized for not addressing secondary reinforcers (such as money), it is still a
useful general framework for this discussion because it provides a context for un-
derstanding the role of stress in reinforcing thought processes.
When an organism’s equilibrium is disrupted, the distance between the current
state and goal state increases, which causes stress. Stress places an organism into
a heightened state of arousal, which can increase associative learning by causing
connections between neurons to form more easily. This is generally considered
adaptive because it enables more rapid experiential learning for lessons most rele-
vant to survival.
Post-Traumatic Stress Disorder (PTSD)
However, extreme stress due to trauma and the consequent sudden and height-
ened learning can have deleterious effects, such as post-traumatic stress disorder
6
(PTSD). In such cases, the brain states that coincided with the trauma are formed
indelibly. The resultant associations are so strong that if those brain states or simi-
lar ones are reproduced by other means (external or internal), it can trigger an as-
sociation to the traumatic event that stimulates a stress response comparable to the
original experience. This stress response then further reinforces the association of
the trigger event to the traumatic event and may even cause new associations to
form that are unrelated to the original event. This self-perpetuating cycle creates
an associative “gravity well” that can eventually link so many aspects of daily ex-
perience to stress that a person becomes effectively paralyzed by anxiety.
Consider, for example, a person who is mugged at gunpoint by someone wear-
ing a ski mask. Subsequently, when the victim is approached in a new context by
an actual skier wearing a ski mask that resembles the one worn by the mugger, he
experiences tremendous anxiety. And since the post-traumatic anxiety is experi-
enced in the novel context of ski slopes, the victim creates new stress associations
to that context and, consequently, avoids skiing.
Fear Circuits
The networks of association between brain states and stress response are some-
times referred to as “fear circuits”. These may be phylogenetic in origin, such as
the innate fear of seeing one’s own blood, or ontogenetic, such as the learned fear
of hospitals (Bracha, 2006). Importantly, these associations are generated by and
apply to perceptual states, regardless of whether they correspond to an external
world state, a dream state, or wakeful thought processes. Indeed, the Ancient
Greek philosopher Epictetus (2004) made the prescient observation that “what
bothers people is not what happens, but what they think of it.”3
Obsessive-Compulsive Disorder (OCD)
While PTSD has a multi-faceted clinical presentation, including such symp-
toms as blackouts (memory loss), it is the manifestation of persistent, intrusive
thoughts, or “obsessions” that is most germane to this discussion. Obsessive-
compulsive disorder4 (OCD) is often the diagnosis given to the presentation of
such chronic rumination. The same notion of fear circuits attributed to PTSD ap-
3 Special thanks to Ernesto Michelucci for re-popularizing this simple quote,
which has deep implications for the human condition.
4 Not to be confused with clinical perfectionism, which is sometimes referred to
as obsessive-compulsive personality disorder (OCPD).
7
plies also to OCD, but does not require a traumatic precursor event, and may in-
volve more abstract concepts.
For example, the perception that a country is moving toward civil war could
generate anxiety in a person that leads to obsession. The increased level of stress
induces hyper-associative learning, such that any new thoughts would be more
likely to be connected to the concept of civil war. For example, a typical shortage
of food at the grocery store could be misconstrued as a sign of stockpiling in antic-
ipation of a food shortage due to war.5 This might link any food-related concepts
to civil war such that any future meal preparation would activate the civil war fear
circuit. Furthermore, food preparation itself would be incorporated into an ex-
panding and self-reinforcing civil war fear circuit.
Treating OCD in Persons
Today, there are two accepted treatments for OCD: a behavioral treatment
called cognitive-behavioral therapy 6(CBT) and pharmacological treatment. Here-
in, we focus on the latter. The most effective drugs for treating OCD are serotonin
reuptake inhibitors (Simpson, 2010), suggesting that a serotonin deficiency may
be responsible for obsessive behavior. Serotonin is a neurotransmitter, a chemical
messenger that supports communication among neurons in the brain. What is
most relevant here is that these drugs are used in the treatment of all anxiety dis-
orders, not just OCD.
Though the specific mechanism by which serotonin alleviates OCD symptoms
is not well-understood, it is the conjecture of this author that reducing anxiety at-
tenuates the hyper-associative growth and reinforcement of fear circuits, thereby
disrupting the rumination cycle. Without constant reinforcement, the fear circuits
diminish over time at a normal rate of memorial decay. This interpretation sug-
gests a computational proxy, discussed below, that might be effective for treating
obsession in distributed cognitive systems.
5 Indeed, according to the Batesonian model described above, this could be de-
scribed as simply another example of context misinterpretation.
6 CBT involves overt associative remapping via exposure and response preven-
tion.
8
A Problem-Solving Superorganism
A superorganism in its most general definition is simply an organism consisting
of many organisms. The present discussion, however, is interested in superorgan-
isms consisting of a technology-mediated collective of human (and possibly ma-
chine) agents, functioning collectively as a distributed information processing sys-
tem. We also assume for this discussion that this system, like all organisms, seeks
homeostatis. Thus, it has drives related to maintaining an equilibrium state and
engages in goal-directed behaviors resulting from those drives. One example of
such a system would be a massively distributed problem solving (Michelucci,
2009) system in which very large numbers of people contribute to solving com-
plex problems that exist in the real world (see Greene and Young, this volume).
Let’s further consider that communication among humans in this superorgan-
ism is mediated by a software-based workflow or cognitive architecture. Thus, the
quality and quantity of information that flows among information processing
agents is both monitored and influenced by the automated control system. Pre-
suming the control system more heavily weights factors that lead most directly to
a solution state, associations relevant to those factors would be reinforced most
heavily.
Obsession in Superorganisms
So how would OCD manifest in such a superorganism. Consider that in such a
problem-solving system “stress” would be characterized by a systemic assessment
of distance between the current state and solution state for whatever problems are
being addressed. For example, if the solution state is a stable earth climate, then
the level of stress in the superorganism might correspond to the perceived distance
between the current state and solution state. Thus, if agents within the system
perpetuate the belief that industrial carbon dioxide emissions are causing climate
change, the system would strengthen the association of carbon dioxide emissions
to stress, leading to increased activity around solving the sub-problem of carbon
dioxide emissions.
It is easy to imagine how such an association could then lead to further associa-
tions to carbon-dioxide emission such as human respiration, which itself could
subsequently lead to the more general observation that all animal respiration adds
the “stress” of climate change. While this association may be valid in terms of
first-order effects, it would ignore systemic effects. And if the associations were
made too strongly due to the perceived influence of carbon dioxide emissions on
climate change-related stress in the system, the problem-solving resources of the
system could become pathologically overcommitted to resolving that carbon diox-
9
ide sub-problem. In other words, the system could be obsessed with reducing car-
bon dioxide emissions to the negligence of a more holistic solution that takes a
more balanced view of the multifaceted nature of the problem.
Treating OCD in Superorganisms
Indeed, such group-based obsession occurs also in natural social systems, as
described in Part I, though in such cases, the only recourse may be behavioral –
that is, policy-based. Engineered systems, however, afford a recourse that might
not otherwise exist. Access to the controlling software would make it possible to
both observe and adjust the rules that govern the strengths of associations among
individual agents in the distributed problem solving system. Decreasing the extent
to which systemic stress influences collaborative activities among agents could
help restore balance to distributed thought processes. Indeed, it is conceivable that
just as with humans, whereby minor changes in neurotransmitter levels can give
rise to significant changes in behavior, small calibrations to parameters that gov-
ern association strength in distributed problem solving algorithms could resolve
obsessive behavior in superorganisms. Whether agents within the system acting
would make such calibrations as implementers of an executive function, by a
completely automated homeostatic algorithm, or by some external “superorganism
psychiatrist” depends upon the evolution of these systems and the co-development
of treatment models.
When Superorganisms are not Organisms
We should not ignore the possibility that other sorts of pathology may exist in
superorganisms that don’t in humans because superorganisms are fundamentally
different than humans – they are themselves composed of interconnected highly
complex organisms. Indeed, superorganisms are a different logical type than hu-
mans. A superorganism is a system of complex systems, which could give rise to
entirely new and unprecedented classes of behavior dysfunction. Since we do not
know what to look for, we may not at first be cognizant of the emergence of such
dysfunction. And once we do become aware, we may need to develop new treat-
ment models and methods specific to those needs. Given the potential impact of
such dysfunction, it would behoove us to minimize the potential for disruptive
surprise by developing our understanding of superorganismic behavioral patholo-
gy in close parallel with the development of superorganisms themselves.
10
Conclusion
The reader may or may not accept various portions of the above speculation, or
may consider the discussion far too incomplete in its presentation for serious con-
sideration. The point we hope will nevertheless be of interest, however, is that at
an individual level, mental pathology can be seen to result from patterned defects
in communication and learning; and that similar defects within a culture or future
engineered social system may result in similar behavioral patterns at the larger
group level. Indeed, the latter may contribute significantly to the former.
If true, one may aspire to develop means to identify and diagnose such infor-
mation processing defects; and to develop interventions to prevent, minimize, or
eliminate the defects or their symptoms.
Acknowledgments
The authors would like to express their sincere gratitude to Mary Catherine
Bateson for her penetrating observations and practicable feedback, all of which
materially improved the conceptual exposition of this chapter.
References
Bateson, G. (1979). Mind and nature: a necessary unity. Cresskill, N.J.: Hampton Press.
Bateson, G., Jackson, D. Haley, J., Weakland, J. (1956). Toward a theory of schizophrenia”, Be-
havioral Science, 1(4): 251-254.
Bateson, M. C. (2010). Willing to Learn: Passages of Personal Discovery. Steerforth Press.
Benedict, R. (1934). Patterns of Culture. Houghton Mifflin Harcourt.
Busemeyer, J. R., & Townsend, J. T. (1993). Decision field theory: a dynamic-cognitive ap-
proach to decision making in an uncertain environment. Psychological review, 100(3), 432–
459.
Bracha, H. S. (2006). Human brain evolution and the “Neuroevolutionary Time-depth Principle:”
Implications for the Reclassification of fear-circuitry-related traits in DSM-V and for study-
ing resilience to warzone-related posttraumatic stress disorder. Progress in Neuro-
Psychopharmacology and Biological Psychiatry, 30(5), 827–853.
doi:10.1016/j.pnpbp.2006.01.008
Epictetus. (2004). Discourses. Courier Dover Publications.
11
Greene, K., and Young, T. (2013). Building blocks for collective problem solving. In P. E.
Michelucci (Ed.), Handbook of Human Computation. New York: Springer.
Heylighen, F. (2013). From human computation to the global brain: the self-organization of dis-
tributed intelligence. In P. E. Michelucci (Ed.), Handbook of Human Computation. New
York: Springer.
Michelucci, P. (2009). Massively distributed problem solving. URL:
http://www.dodsbir.net/sitis/archives_display_topic.asp?Bookmark=36687 [accessed 2013
Jul 5] [WebCite Cache: http://www.webcitation.org/6AeZMC8Nv]
Friedman M. How to Cure Health Care. The Public Interest. 2001. URL:
http://www.thepublicinterest.com/archives/2001winter/article1.html [accessed
2005 Nov 21] [WebCite Cache]
Pavlic, T., Pratt, S. (2013). Superorganismic behavior via human computation. In: Michelucci P,
editor. The Handbook of Human Computation. New York: Springer.
Seward, J. (1956). drive, incentive, and reinforcement. Psychological Review, 63, 19-203.
Simpson, H. B. (2010). Pharmacological treatment of obsessive-compulsive disorder. Current
topics in behavioral neurosciences, 2, 527–543.
Whitehead, A. N., & Russell, B. (1927). Principia Mathematica. Cambridge University Press.
