Choice • The Nature of Want • Possibility Space & The Efficient Frontier • Information Scarcity • Exploring Possibility Space • The Free Energy Principle • Multi-Agent Utilities • Conclusion • Works Referenced
Choice
Why do we make choices?
Choices are actions that preclude others. We cannot have our cake and eat it, too.
Whenever we are faced with different mutually exclusive alternatives, we must compare the various options with reference to some standard of evaluation.
This evaluation is called a utility function, and rational behaviour is when we choose the option which maximises utility.
Choice requires knowledge of what we want and what our options are.
Students taking an introductory economics course will no doubt hear that economics deals with decision-making under scarcity. Human desires are supposedly unlimited and resources are supposedly scarce.
If these are the two starting points of our economic theory, it seems important to ensure they hold in situations where we typically apply economic reasoning.
Scarcity is the foundation of the essential problem of economics: the allocation of limited means to fulfill unlimited wants and needs.
Jim Chappelow, 2019, Scarcity
Lionel Robbins (1932) phrases economic choice as a dialectic between these two assumptions. Problems of what we can do are for technologists; problems of what we should do are for philosophers.
Economics is important, says Robbins, because it describes the interaction of possibilities and desires.
Therefore he has to choose. He has to economise. Whether he chooses with deliberation or not, his behaviour has the form of choice. The disposition of his time and his resources has a relationship to his system of wants. It has an economic aspect.
Lionel Robbins, 1932, Essay on the Nature and Significance of Economic Science
Furthermore, Paul Krugman and Robin Wells (2009) note that individual choice alone is not enough to understand the function of economies.
Interactions between choices across agents (as described by game theory and decision theory), across time (as described by temporal discounting and multi-stage game theory), and across situations (as challenged by the intransitive preferences observed in behavioural economics) are also described by economic models.
Krugman and Wells’ textbook Economics begins with four key assumptions about individual choice.
- Resources are scarce – as they put it, “you can’t always get what you want”
- The real cost of something is what you give up to get it – rational choices minimise opportunity cost
- “How much?” is a decision at the margin – rational choices cannot be improved by small changes
- People usually exploit opportunities to make themselves better off – people’s choices are influenced by incentives
This framework takes means and ends as fixed and known. This is consistent with Robbins’ claim that economics concerns the interaction between the two.
However, describing the nature of this interaction requires an understanding of both sides of the framework: desires (ends) and possibilities (means).
The Nature of Want
Various metaphysical and epistemic problems arise when considering values and desires.1 What is it that we want? How do we find out what we want?
Nirvana is not necessarily single bliss. It is the complete satisfaction of all requirements.
Lionel Robbins, 1932, Essay on the Nature and Significance of Economic Science
Robbins divides wants into two broad categories: real income and leisure.
Time is the main resource in this model, and Robbins says it is too short to fully satisfy either desire. We must choose between possible mixes of the two categories.
The choice of words here is somewhat lacking. “Real income and leisure” might be uncharitably reformulated as “greed and sloth”, which are certainly not the only predictors of human satisfaction.
Daniel Pink (2009) identifies autonomy, mastery and purpose as the three contemporary dimensions of individual motivation.
This may be useful for managers and organisational psychologists, but a description of complex, interconnected economic systems requires a more quantitative model of behaviour.
Computing the predictions of qualitative models generally becomes intractable at the scale of a firm or a country.
Also, humans are not the only agents considered by contemporary economists: the concept of legal personhood is at least as old as Ancient Rome, and is formalised by the economic theory of the firm.
A more general formulation of Robbins’ dichotomy is prevalent in discourse surrounding Artificial General Intelligence: the instrumental/terminal value distinction.
Some desires, such as the desire for money, are said to be instrumental because their existence is dependent on other, supposedly more fundamental desires.2 This corresponds to Robbins’ real income desires.3
Desires not dependent on others are said to be terminal: they exist regardless of an agent’s environmental context, and do not serve any greater purpose than themselves. In Robbins’ framework, these are the leisure desires.4
Rapoport (1953) notes that shared cultural values have a tendency to begin as conditional, instrumental goals, but become increasingly terminal over time.
The progressive-traditionalist culture war can be seen as a conflict over the appropriateness of heuristics in evaluating the utility of alternatives, especially in a changing context.5
Even in the case of true terminal goals, it is often difficult to figure out what we truly want.
According to psychologist William Berry (2017), “research indicates humans are bad at predicting what will make them happy”.
In any case, leading a meaningful life is a more complex problem than maximising a simple aggregate of moment-to-moment happiness.
It is the very pursuit of happiness that thwarts happiness.
Viktor Frankl
Buddhist, Epicurean and Stoic traditions alike all have some notion of removing desires through will, not action – though as Thorn (2020) observes, a state of desirelessness “just sounds like… depression”.
Desirelessness is, in a sense, the opposite of what it means to be human, or even alive.
It is conceivable that living creatures may exist whose “ends” are so limited that all goods for them are “free” goods, that no goods have specific significance.
Lionel Robbins, 1932, Essay on the Nature and Significance of Economic Science
Is choice even possible if we don’t know what we want?
Berry’s observation implies that our future selves have uncertain utility functions.
Expectations about future utility are a combination of beliefs about the world and beliefs about ourselves.
Choices depend in part on a model of the unknown internal state of our brain and how it corresponds to our qualitative assessment of satisfaction.
Even with a known utility function, choice still requires knowledge of what is possible. Economists usually phrase this in terms of resources.
1Metaphysics is primarily concerned with the fundamental nature of that which exists; epistemology is primarily concerned with how we obtain and justify claims about what exists and what it is like.
2The possibility that desire dependencies are circular is not considered. Popper’s principle of reflexivity may also be relevant here.
3As Clint Eastwood put it: “Man’s gotta make a living”.
4This is reminiscent of Aquinas’ idea of necessary and contingent beings in the context of theology and cosmological ontology.
5Nieztsche wrote of a “transvaluation of all values”; see also Wes Cecil’s lecture series of the same name.
Possibility Space & The Efficient Frontier
Why do individuals have to make choices? The ultimate reason is that resources are scarce.
Krugman & Wells, 2009, Economics
Krugman & Wells define resources as “anything that can be used to produce something else”.
Production is generally an instrumental goal: its value is dependent on the output of the production process.
They list four key economic resources:
- Land
- Labor – the time of workers
- Capital – machinery, buildings, and other man-made productive assets
- Human capital – the educational achievements and skills of workers
They provide a further list of scarce resources:
- Natural resources – resources that come from the physical environment, such as minerals, lumber, and petroleum
- Human resources – labor, skill, and intelligence.
- Clean air and water, in the context of excessive population growth
A resource is scarce when there’s not enough of the resource available to satisfy all the various ways a society wants to use it. There are many scarce resources.
Krugman & Wells, 2009, Economics
It is notable that these two concepts are separated. Not all resources are scarce in the same way, and many resources have some component of abundance.
A single piece of information can be replicated relatively cheaply. There is usually little competition over sunlight outside of urban environments. There is no shortage of cat videos on the internet.
Save in very special circumstances, the fact that we need air imposes no sacrifice of time or resources. The loss of one cubic foot of air implies no sacrifice of alternatives. Units of air have no specific significance for conduct.
Lionel Robbins, 1932, Essay on the Nature and Significance of Economic Science
Nonetheless, there are usually limitations on the productive capacity of resources, which limit the satisfaction of terminal desires.
Such limitations are generally expressed in terms of a production possibility space: of all the conceivable outputs of production, only some are possible.
The material means of achieving ends are limited. We have been turned out of Paradise. We have neither eternal life nor unlimited means of gratification. Everywhere we turn, if we choose one thing we must relinquish others which, in different circumstances, we would wish not to have relinquished. Scarcity of means to satisfy given ends is an almost ubiquitous condition of human behaviour.
Lionel Robbins, 1932, Essay on the Nature and Significance of Economic Science
If we suppose that more of a particular good is better than less, all else being equal, then we can define a subset of possibilities which are efficient: they involve the maximum quantity of some particular good for given quantities of all others.
Any point within the efficient frontier represents a particular attainable production mix.
Image Courtesy of Everlong, GFDL & CC-BY-SA 2.5
Alex Lenail (2014) notes that technological innovations like computers and robots change the nature of economic scarcity.
The automation of increasingly complex business processes, a hierarchical view of instrumental economic goals, and accelerating rates of change increase the productive capacity of the economy and reduce the need for human input.
In high school econ, we were told on the first day that economics was the study of models of resource-distribution in a world of scarcity. Of infinite desires, and finite resources.
Alex Lenail, 2014, Infinite wants, infinite resources
Automation does not, however, entail unlimited possibility.
Even automated processes require physical inputs, time, and often maintenance.
A truly post-scarcity economy must be founded on perfectly sustainable, fully automated production processes, which are impossible in practice.
Even then, there is still the question of who has access to the results of production: this forms the basis of the classic Marxian objection to capital accumulation.
So long as there are scarce resources to bargain for, or economic values lag behind productive capacity, artificial scarcity will still be imposed.6
Desires are satisfied by action. Action utilises available resources (often to produce new resources), and so the availability of primary resources – tangible or intangible – is what limits the satisfaction of desires.
Human agents are the fundamental means and ends of economic activity. People combine labour with capital to create value, and all resource consumption by firms is, ideally, a means to resource consumption by individuals.
Whereas technological problems entail finding the most efficacious means of accomplishing a given end, economic problems entail allocating scarce means to multiple and competing ends.
Martin & Petersen, 2019, Poverty Alleviation as an Economic Problem
With a scarcity mindset, all economic choices are allocative.
We may move our production point towards or along the efficient frontier.
Supply-side economic growth, on the other hand, moves the frontier itself: technological advances or population growth can change what is attainable and allow greater satisfaction of desires.
Changes to the possibility frontier can alleviate the problem of scarcity itself.
The technological innovations of the past few decades have challenged established economic paradigms and forced a reconsideration of scarcity as a defining characteristic of choice.
Even the scarcity of time is being challenged by increases in life expectancy which may eventually outpace aging itself.
Nonetheless, a post-scarcity economy continues to seem more of an ideal direction than an attainable goal.
What computers and their programmers have achieved in the past half-century or so is the construction of schema that render any well-defined pattern or procedure into an automatic routine that can be executed an indefinite number of times. Put simply: once a job has been programmed, it never needs to be done by anyone ever again.
Alex Lenail, 2014, Infinite wants, infinite resources
6Classical and Keynesian economics would predict deflation as a result of widespread automation, which will, paradoxically, reduce consumption. Faced with a choice between buying now or buying later at a lower price, rational consumers will be more likely to defer purchases that are not urgent.
Information Scarcity
Data is certainly not scarce in the way most natural resources are.
Any one piece of data can be duplicated and transmitted quickly and cheaply.
Nonetheless, storage space, transfer speed, and the variety and quality of data available is still limited.
The unique nature of data scarcity has led to a mixture of traditional data markets competing with freely available datasets.
Information in a business context is value derived from data.
Data must be processed and analysed to gain relevance and provide actionable insights – this is the value proposition of data science.
Brains and machines alike have limited processing power. We must decide where to focus our attention and how to interpret the data we take in.
Copyright regulation and paywalls introduce artificial scarcity into the market for information. Neither data nor information are, as is sometimes claimed, public goods.7
The supply of information and its use in economic decision-making was first addressed by Friedrich Hayek in 1945, and has since been explored from a variety of economic perspectives.
Here we will focus primarily on its relationship to value and utility.
7In economics, a public good is one which agents cannot exclude each other from using, and where use by one agent does not inhibit use by another. Public goods can be consumed by a theoretically unlimited number of people, but the variety of public goods produced is limited, and “pure” public goods don’t really exist in practice. While the second criterion (non-rivalry) is generally true of data and information, they are clearly excludable goods, and so not public goods.
Exploring Possibility Space
Choice is the interaction between possibility and desire. How do we know what is possible?
Exploring possibility space is surprisingly difficult.
A full appreciation of everything that is possible would require comprehensive understanding of every academic discipline, some of which do not yet exist.
Common problems in appreciating the breadth of possibility space include functional fixedness (an inability to consider possible atypical uses of a tool) and learned helplessness (a false belief that something which was previously uncontrollable remains uncontrollable).
Lateral thinking is necessary to see the full range of possible choices in complex scenarios. Information about the nature and variety of these options allows us to make choices which approach the efficient frontier of possibility space.
Information about the results of our choices provides better approximations for our utility functions across the known possibility space by evaluating choices with reference to instrumental or terminal goals.
This is a common application of statistics and data science, but similar techniques may allow us to better explore and parameterise the whole space of possibilities.
Information management is rapidly becoming a key organisational function.
Effective information management allows organisations to explore possibility space, better approximate their utility function for each possibility, and understand what the instrumental (or even terminal) goals of the organisation truly are.
Managers must also understand how the interaction of individual utility functions with organisational incentive structures results in choices that harm or hinder the organisation, and how the organisational utility function emerges from and relates to those of its members or stakeholders/customers.
Organisations must address information silos and ensure business processes allow for efficient and effective information flows. Information systems should be designed and automated with organisational goals and values in mind.
This is more than a concern about websites or databases. Effective managers or anyone engaged in professional communication need to convey relevant information and motivate appropriate action, and effective semantics and communication are essential for defining roles, communicating needs and values, and evaluating organisational culture.
The Free Energy Principle
Here we are, sentient creatures with bundles of desires and aspirations, with masses of instinctive tendencies all urging us in different ways to action.
Lionel Robbins, 1932, Essay on the Nature and Significance of Economic Science
Karl Friston’s Free Energy Principle is a mathematical formulation of the relationship between action and information.8
The action of physical systems always involves some quantity which is maximised or minimised.
A thermostat minimises the difference between the desired and actual temperature. A fish minimises the distance between it and its food, while maximising its distance from the nearest predator. A rational economic agent acts to maximise their expected utility function.
The action of any system can be interpreted as a combination of external beliefs (internal states corresponding to external realities) and a internal beliefs (internal states spurring action to change external reality).
Friston interprets the quantity maximised by systems as something approximating probability, so that an agent either acts or changes its internal states to increase its certainty about the world and minimise the frequency of surprising events.
“Value” is that which one acts to gain and keep, “virtue” is the action by which one gains and keeps it. “Value” presupposes an answer to the question: of value to whom and for what? “Value” presupposes a standard, a purpose and the necessity of action in the face of an alternative. Where there are no alternatives, no values are possible.
Ayn Rand, 1961, For the New Intellectual
The free energy principle suggests an interesting operational definition of utility – that which is maximised when an agent has correct, certain beliefs about the external world.
Everything we do is exactly what we desire at the moment of decision, given our beliefs about our environment.
This implies that the impossible cannot be desired – all desires lie within the known possibility space.
Perhaps I desire a pizza in the next 10 seconds. This is certainly possible if there is a pizza right next to me. Desiring the pizza in the next 5 seconds is probably possible too, but it is unlikely I will be able to get the pizza within the next 500 milliseconds.
As my timeframe gets smaller, the physical effort associated with fetching the pizza in the desired timeframe grows. I cannot, given this additional effort, desire the pizza in the next millisecond – such a process would be likely to cause me undesirable physical injury.
As we approach the frontier of possibility, other costs conspire to prevent the impossible from being truly desirable.
Our utility function is implicit in our actions and beliefs. However, people often report acting against their wishes. How are we to interpret this?
It is possible that in some cases such reports are outright lies. A more charitable interpretation is that we are equivocating two different definitions of desire.
Our action is a particular aggregate of the costs and benefits associated with each possibility.
Anything we do can be interpreted as maximising utility, but our conscious brain won’t necessarily identify with all of the desires we act to satisfy, resulting in a mismatch between what we do and what we say we want.
Thinking the words “I want to lift my left hand” will not result in the lifting of your left hand without the cooperation of both deliberate and automatic cognitive processes.
Which is the “true” sum of our desires? Both interpretations are important for solving economic problems. What we say we want is often more accessible than our revealed preferences.
8The free energy principle comes from mathematical neuroscience, and is more complex and nuanced than the summary given here. See Buckley et. al. (2017) for an in-depth mathematical appraisal of Friston’s work, and Alexander (2018) for a more accessible explanation.
Multi-Agent Utilities
The problem becomes even more complex when aggregating across multiple actors.
Bentham’s utilitarian calculus results in numerous ambiguities on the level of populations, such as Parfit’s mere addition paradox.
For any perfectly equal population with very high positive welfare, there is a population with very low positive welfare which is better, other things being equal.
Derek Parfit, 1984, Reasons and Persons
This summation problem is a key concern in negotiating fair political, economic and organisational arrangements – as Krugman & Wells put it, economists must understand the “interaction between individual choices”.
Operationally, our desires are finite – but our conscious mind may certainly claim to desire more than is possible.
Conclusion
Choice is a resolution of complex interactions between means and ends.
The freedom to choose, and the resulting problem of correct choice, are key existential concerns.
Man is condemned to be free; because once thrown into the world, he is responsible for everything he does.
Jean-Paul Sartre, 1956, Being and Nothingness
The interaction of choices across individuals and across time is a key concern of politics, sociology and economics, and deliberate solution of this problem necessitates economic thinking.
However, changes in the condition of humanity will necessitate changes in our the nature of our economic frameworks.
Works Referenced
Scott Alexander (2018), God help us, let’s try to understand Friston on free energy, Slate Star Codex
Thomas Aquinas (1485), Summa theologica
William Berry (2017), You don’t know what you want, Psychology Today
Christopher L. Buckley et. al (2017), The free energy principle for action and perception: A mathematical review, Journal of Mathematical Psychology
Jim Chapelow (2019), Scarcity, Investopedia
Viktor Frankl (1946), Man’s search for meaning
Friedrich Hayek (1945), The use of knowledge in society, The American Economic Review
Paul Krugman and Robin Wells (2009), Economics: Second edition
Alex Lenail (2014), Infinite wants, infinite resources, Eclectic Esoterica
Adam Martin and Matias Petersen (2018), Poverty alleviation as an economic problem, Cambridge Journal of Economics
Derek Parfit (1984), Reasons and persons
Daniel Pink (2009), Drive: The surprising truth about what motivates us
Ayn Rand (1961), For the new intellectual
Anatole Rapoport (1953), Operational philosophy
Lionel Robbins (1932), An essay on the nature and significance of economic science
Jean-Paul Sartre (1956), Being and nothingness
Oliver Thorn (2020), Beauty in ugly times