Abstract
A formal model for the statement of
normative moral theories (NMTs) is proposed as a tool for the facilitation and
expansion of academic discourse on ethics and meta-ethics. The model is argued
to bring forth hidden assumptions, eliminate abstractions and facilitate
understanding, to enable synergies with other sciences in clearly defined areas
and to set clear boundaries for the scope of peer review.
Preface
Conscious systems posit Normative Moral
Theories to guide their actions.
The universe is a closed system which
contains conscious beings. Conscious beings are defined as the subsystems with at
least the following characteristics:
·
Analytical capability. They are able to
decompose the perceived universe into clearly defined, layered classes of
structures, with clearly defined states.
·
Synthetic capability. They are able to
formulate general statements to describe relationships between the classified
structures.
·
Sense of identity. They have the
perception of a self, which they classify as distinct from the rest of the
universe.
·
Planning capability. They have the
perception of time, can identify cause and effect and formulate plans of
action.
·
Volition and Action. They have innate drives,
the perception of will and the ability to execute automated and planned actions
to achieve their goals.
The innate drives bias the conscious systems
towards particular goals and actions. However, once fundamental needs and wants
are satisfied, volition is quite malleable and amenable to cognitive influences.
Due to the aforementioned characteristics, conscious systems perceive
themselves as having free will and proceed to set additional goals, not
directly dictated by or even contradictory to their innate drives. The
formulation of a Normative Moral Theory is one such expression of the
perception of free will.
Benefits of a formal model for NMTs
Bring
forth hidden assumptions, eliminate abstractions and facilitate understanding. Set
boundaries for criticism.
NMTs have a long tradition of providing
guidelines based on the assumptions and knowledge of their time. As a result,
there was little need for rigorous justification of the particular guidelines.
In free, democratic societies it is accepted that individuals may hold
different beliefs, as long as their actions do not stray considerably from the
range of actions accepted by the society. At times, the wide range of opinions
means that no true consensus can ever be reached on particular subjects,
leading to constant conflicts and power struggles. In an ideal situation,
honest discourse may lead to clear understanding of the reasoning behind the
various beliefs and to the attainment of a relatively fair compromise. However,
as long as the assumptions are not unambiguously formulated and the limits of
criticism are not well established, misunderstandings and aphorisms are
unavoidable.
A formal model for NMTs can facilitate
understanding by restating implicit assumptions and the reasoning used to reach
particular conclusions, in widely understood terms. A formal model can expose
fundamental, irreconcilable differences in viewpoints that need to be accepted,
as well as logical fallacies that can be constructively criticized. A formal
model can’t eliminate conflicts, but it can offer a framework for more
constructive dialog between the opposing factions.
Enable
synergies with social and natural sciences, mathematics and computer science.
Ethics is certainly not a science and
can never become one. Conscious beings may agree on many things about their
world, but they will always disagree on how one should lead one’s life.
However, a formal model can permit ethicists to utilize the considerable
arsenal built for other sciences, in order to facilitate predictions about the
effects of certain actions on the pursued goals. Science can only help when the
problem is clearly formulated and ethical issues never are. We will later show
how a formal model enables the use of software simulations, statistics and
chaos theory to rigorously justify prescribed courses of action, based on
arbitrary assumptions.
Clearly
outline the boundaries between ethics and science.
One practical application of a formal model is to once and for all
define what science can and what it can’t say about how one should lead one’s
life. The subject has recently become of interest once more, as the ideas of
people such as XXX and YYY are
appealing to many.
Elements of a NMT
All NMTs assume or define the
relationship between the individual and the rest of the world (describe what
is), set goals (posit oughts) and prescribe ways to achieve the goals.
NMTs
describe what is
·
A formal NMT MUST explicitly describe all perceived
classes, states and interrelationships, in terms unambiguously understood by a
fluent speaker of the language.
To posit an NMT, conscious beings either
assume as common knowledge or define the relationship between the individual
and one or more other structures. They arbitrarily decompose the perceived
closed system (universe) into subclasses (e.g. family, society, environment
etc.), identify class instance states (e.g. happy parent, just society) and define
interactions between the individual and the instances of such classes (e.g. a
human depends on plants and animals for sustenance). Defining a class requires
setting arbitrary, but clear boundaries. For instance, a conscious system may
split the universe into just two classes, one containing the conscious system
itself and the other containing everything else.
The subjectivity of perception and the
practical difficulties of accurately defining abstract states such as ‘just’ or
‘happy’ guarantee that the consensus on what is can never be universal. As a result:
NMTs
posit oughts
·
A formal NMT MUST assign arbitrary value functions to
class instances and/or instance states.
·
A formal NMT MUST mathematically define targets for
the assigned value functions
NMTs set goals for the individual, the
collection of individuals or any other instance of the described or assumed
classes. They assign arbitrary values to instances of the described classes
(e.g. an individual life), or to perceived instance states (e.g. an
individual’s happiness).
To posit clear goals, the NMT must assign non-zero value to instances of
at least one class. The value may well be infinite, suggesting that no system
instance can be sacrificed to pursue another goal. A NMT may also assign values
to particular system instance states. The NMT must then set a target for the assigned values. For example, a
NMT might assign a constant value (V) to every human life L and another value
to the human state of happiness, for each human H(i). It might then set various
goals (G), such as the following:
NMT 1: As many people as possible, as
happy as possible
G = ∑_(i=1)->L〖(H(i)+V〗)
NMT 2:
As many people as possible, as happy as possible, but with a lowest
bound and a greater emphasis on happiness
G = ∑_(i=1)->L〖(H(i)*V〗) where H(i) > C
NMT 3:
Seek to maximize the average happiness
G =(∑_1->L〖(H(i)+V)〗)/L
NMT 4: Look into the long-term and
target a sustainable level of happy population (Expected value, for time 0 to
infinity)
The assigned values themselves need not
be constants. For instance, the value may be a function of the number of
instances of the particular system, allowing for increasing value as the number
of instances decreases (endangered species).
NMTs
prescribe actions to achieve the set goals
·
A formal NMT MUST provide rigorously justified, clear
guidelines as to how the targets should be pursued, under particular
circumstances.
·
A formal NMT MAY provide rigorously justified,
practical rules of thumb to guide decision making in a wide range of
situations.
A NMT prescribes actions and methods of
choosing between actions that lead to the attainment of the set target(s). Even
NMTs that agree both on what is and what ought to be, do not necessarily reach
the same conclusions as to how one should act. Even with the simplistic goals
described above, accurate value determination would be practically impossible,
especially for large time scales. As a result, most conscious systems (humans
for certain) would need to resort to heuristics, reasonable assumptions and
simulations, based on largely acceptable theories and widely held beliefs. A formal NMT must therefore justify its
prescriptions based not only on what is and what ought to be, but also on the
methodology used to reach the conclusion.
The effects of certain actions may
depend on how prevalent they are (e.g. free rider problem). More generally,
certain actions may not be directly or even indirectly detrimental to the
pursuit of the set goals, but they may be shown to increase the risk of straying
from the desired path. To cover such cases, a NMT may prescribe practical rules
of thumb, such as the universal law principle, taking into consideration the
effect on the set goals, if a particular action became a universal law. The
justification of such wide-ranging guidelines can’t be expected to be complete,
but a NMT must show how their application generally leads to the desired goals.
The boundaries of criticism
Completeness
A formal NMT
is complete when it describes all relevant system classes and interconnections.
Formal NMTs need not agree on what is, but they do
need to provide a complete view of the proposed network of systems. An obvious
point of contention is the proposition of a ‘Deity’ class, with one or more
instances. An NMT can’t be criticized for proposing that such a class does
exist, but it can be criticized if it fails to completely describe the
interactions between the instance(s) of such a class and the rest of the
described network. As a result, religions supporting that “we can’t know the
will of God” would be quite difficult, if not impossible to restate as
complete, formal NMTs. To offer a less contentious example, a NMT can be shown
to be incomplete, if it fails to account for the effect of humanity’s actions
on the environment and the feedback effects of those actions to humanity’s
well-being in large timescales.
The number of system classes defined by
a formal NMT is bounded only by the conscious system’s capacity to accurately
describe the resulting network. A fine separation carries the risk of misclassifying
poorly understood systems or poorly addressing the dependencies between two
identified classes, therefore rendering the moral theory incomplete.
Consistency
A NMT is
consistent when its goals are attainable and when its prescriptions can be
proven to serve its goals.
When positing oughts, the value functions and
targets must be carefully chosen, in order for the goals to be theoretically
attainable. A NMT may obviously posit as many targets as it wishes, at the
peril of rendering the consistency requirement practically impossible, since the
complexity of the systems under consideration will probably lead to conflicting
goals. Conflicting goals will need to be restated
in a way that provides a single, clear target state, which will involve a
compromise between the unattainable ideals. To provide another example,
infinite values can’t be assigned to instances of all system classes, because
the second law of thermodynamics guarantees that not all system structures can
be preserved across time.
NMTs are supposed to prescribe actions
that protect system instances and system states with non-zero values, taking
into account all system to system interconnections. Regardless of the
prescribed method for deciding on a course of action, the resulting actions
must be shown to be consistent with the goals. A critic of the NMT may challenge a rule of thumb or
heuristic used, by presenting a case where the prescribed action undermines the
NMT’s own goals.
Rigour
A formal NMT is rigorous when it can
make testable hypotheses for the effect of certain actions on the desired
goals.
We explained how a superficial heuristic
may lead to the violation of the consistency requirement. Lack of rigor is the
greatest threat for a formal NMT. The scientific method is based on hypotheses,
predictions and independently repeatable results. Given the complexity of the
systems under consideration, proposing a fully testable NMT may be quite
difficult, but not theoretically impossible. The less a NMT relies on
heuristics, the more room it leaves for rigorous use of mathematics (e.g. chaos
theory, game theory, statistics, network theory), peer reviewed socioeconomic
and anthropological studies, computer simulations etc. If there is one easy
criticism one could make for all informal NMTs is their lack of rigor. The path
towards rigorous formal NMTs will be lengthy and arduous, but the benefits
outweigh the costs.
A formal NMT can be criticized for lack
of rigor when it describes what is, if it fails to provide unambiguous
definitions of the described classes, or if it provides unsubstantiated claims on
the dependencies between the classes.
A
formal NMT can be criticized for lack of rigor when it posits oughts, if the
value functions it provides are not measurable. For instance, ‘happiness’ is
not a clearly defined system state, but dopamine and serotonin levels are.
A formal NMT can be criticized for lack of rigor when it prescribes actions, if its
reasoning contains logical fallacies, or if it makes poor use of scientific
methodologies. For instance, a NMT might utilize a computer simulation to predict
the effect of a certain action on a set goal, but the software may later be
found to have had a critical bug that skewed the results.
The
boundaries between ethics and science
The role of science in describing what
is
·
Science can
assist in identifying potential elementary system classes (e.g. wolves vs dogs)
and especially in describing the interconnections between the various system
classes.
·
Science has
no say on whether a NMT treats all animals as a single superclass or whether it
separates them into dogs and non-dogs.
The role of science in positing oughts
·
Science can
assist in clearly defining system states (term disambiguation) and in
calculating value functions.
·
Science has
no say in the values assigned or in the set targets.
The role of science in prescribing
actions
·
Science can
assist in the complex predictions of the effects of particular actions. Aside
from providing tools to facilitate difficult calculations, it can be used to
prove that a utilized heuristic is unfounded because certain feedback effects
from particular interconnections were ignored during the calculations.
Innate Boundaries
The
definition of a formal NMT which was provided above seemingly permits a
boundless set of possibilities. However, we should emphasize that the
perceptions of conscious entities are heavily biased by the innate capabilities
described in the preface. The more innate characteristics two conscious
entities share, the closer their perceptions of the world and their place in it
will be. For instance, biological entities are generally characterized by a
drive for self-preservation which would not necessarily be present in an
artificial consciousness. The innate selfishness undoubtedly biases the NMTs
posited by biological entities and they generally choose to assign high values
on the instances of their own class. To illustrate with an exception, NMTs
which support the idea that humans should stop reproducing because of their effect
on the environment have been posited (reference) but their reception from the vast majority of humans
has been one of condescending indifference (reference).
Therefore,
it is quite possible for conscious entities with many common characteristics to
reach a consensus on a subset of the possible descriptions of the world and on
at least some of the value functions assigned to the agreed upon classes. One
should always remember though, that the agreed upon ‘truth’ is their own agreed
upon truth, which is always subject to revision from future generations of
entities with slightly different characteristics.