Aaron Sloman, professor of philosophy at the School of Computer Science of the University of Birmingham, counts certainly as one of the most influential theoreticians regarding computer models of emotions. In an article from 1981 titled "Why robots will have emotion" he stated:
Like Bates, Reilly or Elliott, Sloman also represents the broad and shallow approach. For him, it is more important to develop a complete system with little depth than individual modules with much depth. It is his conviction that only in this way a model can be developed which reflects reality to some extent realistically.
Sloman and his coworkers in the Cognition and Affect Project have, since 1981, published a lot of works on the topic "intelligent systems with emotions", which can be divided roughly into three categories:
To understand Sloman's approach correctly, one must see it in the context of his epistemological approach which is not concerned primarily with emotions, but with the construction of intelligent systems.
I shall try to sketch briefly the core thoughts of Sloman's theory because they form the basis for the understanding of the "libidinal computer" developed by Ian Wright (see below).
Sloman's interest lies not primarily in a simulation of the human mind, but in the development of a general "intelligent system", independent from its physical substance. Humans, bonobos, computers and extraterrestial beings are different implementations of such intelligent systems - the underlying construction principles are, however, identical.
Sloman divides the past attempts to develop a theory about the function modes of the human mind (and thus of intelligent systems generally) into three large groups: Semantics-based, phenomena-based and design-based.
Semantics-based approaches analyze how humans describe psychological states and processes, in order to determine implicit meanings which are the basis of the use of words of everyday language. Among them he ranks, among others, the approaches of Ortony, Clore and Collins as well as of Johnson-Laird and Oatley. Sloman's argument against these approaches is: "As a source of information about mental processes such enquiries restrict us to current `common sense´ with all its errors and limitations." (Sloman, 1993, p. 3)
Some philosophers who examine concepts analytically, produce, according to Sloman, semantics-based theories, too. What differentiates them from the psychologists, however, is the fact that they do not concentrate on existing concepts alone, but are often more interested in the quantity of all possible concepts.
Phenomena-based approaches assume that psychological phenomena like "emotion", "motivation" or "consciousness" are already clear and that everybody can intuitively recognize concrete examples of them. They try therefore only to correlate measurable phenomena arising at the same time (e.g. physiological effects, behaviour, environmental characteristics) with the occurrence of such psychological phenomena. These approaches, argues Sloman, can be found particularly with psychologists. His criticism of such approaches is:
Design-based approaches transcend the limits of these two approaches. Sloman refers here expressly to the work of the philosopher Daniel Dennett who essentially shaped the debate around intelligent systems and consciousness.
Dennett differentiates between three approaches if one wants to make forecasts about an entity: physical stance , design stance and intentional stance . The physical stance is "simply the standard laborious method of the physical sciences" (Dennett, 1996, p. 28); the design stance, on the other hand, assumes "that an entity is designed as I suppose it to be, and that it will operate according to that design" (Dennett, 1996, p. 29). The intentional stance which can be regarded according to Dennett also as a"sub-species" of the design stance, predicts the behaviour of an entity, for example of a computer program, "as if it were a rational agent" (Dennett, 1996, p. 31).
Representatives of the design-based approach proceed from the position of an engineer who tries to design a system that produces the phenomena to be explained. However, each design does not require at the same time also a designer:
A design is, strictly taken, nothing else than an abstraction which determines a class of possible instances. It does not have to be necessarily concrete or materially implemented - although its instances can quite have a physical form.
For Sloman, the term "design" is closely linked with the term "niche". A niche is also a not a material entity and no geographical region. Sloman defines it in a broad sense as a collection of requirements to a functioning system.
Regarding the development of intelligent agents in AI, design and niche play a special role. Sloman speaks of design-space and niche-space . A genuinely intelligent system will interact with its environment and will change in the course of its evolution. Thus it moves on a certain trajectory through design-space . With it corresponds a certain trajectory through the nichespace , because through the changes of the system it can occupy new niches:
Sloman identifies different trajectories through the design-space: Individuals who can adapt themselves and change, go through so-called i-trajectories . Evolutionary developments which are possible only over generations of individuals, he calls e-trajectories . And finally there are changes in individuals that are made from the outside (for example debugging software) and which he calls r-trajectories (r for repair).
Together these elements result in dynamic systems which can be implemented in different ways.
For Sloman, one of the most urgent tasks exists in specifying biological terms such as niche, genotype etc.more clearly in order to be able to exactly understand the relations between niches and designs for organisms. This would also be a substantial progress for psychology:
Sloman grants that the requirements of design-based approaches are not trivial. He names five requirements which such an approach should fulfill:
A design-based approach does not necessarily have to be a top-down approach. Sloman believes that models which combine top-down and bottom-up will be most successful.
For Sloman, design-based theories are more effective than other approaches, because:
What a design-based approach sketches, are architectures. Such an architecture describes which states and processes are possible for a system which possesses this architecture.
From the quantity of all possible architectures, Sloman is particularly interested in a certain class: "..."high level" architectures which can provide a systematic non-behavioural conceptual framework for mentality (including emotional states)." (Sloman, 1998a, p. 1) Such a framework for mentality
An architecture for an intelligent system consists, according to Sloman, of four substantial components: several functionally different layers, control states, motivators and filters as well as a global alarm system.
9.2.1. The layers
Sloman postulates that every intelligent sytem possesses three layers:
The reactive layer is the evolutionary oldest, and there is a multitude of organisms which only possess this layer. Schematically, a purely reactive agent presents itself as follows:
Fig. 13: Reactive architecture (Sloman, 1997a, p. 5)
A reactive agent can make neither plans nor develop new structures. It is optimized for special tasks; with new tasks, however, it cannot cope. What it is missing in flexibility, it gains at speed. Since almost all processes are clearly defined, its reaction rate is high. Insects are, according to Sloman, examples for such purely reactive systems, which prove at the same time that the interaction of a number of such agents can produce astonishingly complex results (e.g. termite towers).
A second, phylogenetically younger layer gives an agent more qualities by far. Schematically, this looks as follows:
Fig. 14: Deliberative architecture (Sloman, 1997a, p. 6)
A deliberative agent can re-combine its action repertoire arbitrarily, develop plans and evaluate them before execution. An essential condition for this is a long-term memory in order to store plans not completed yet or to rest and evaluate later the probable consequences of plans.
The construction of such plans proceedes gradually and is therefore not a continuous, but a discrete process. Many of the processes in the deliberative layer are of serial nature and therefore resource-limited. This seriality offers a number of advantages: at any time it is clear to the system which plans have led to a success, and it can assign rewards accordingly; at the same time, the execution of contradicting plans is prevented; communication with the long term storage is to a large extent error free.
Such a resource-limited subsystem is of course highly error-prone. Therefore filtering processes with variable thresholds are necessary, in order to guarantee the working of the system (see below).
The phylogenetically youngest layer of the system is what Sloman calls the meta management:
Fig. 15: Meta management architecture (Sloman, 1997a, p. 7)
This is a mechanism which monitors and evaluates the internal processes of the system. Such a subsystem is necessary to evaluate the plans and strategies developed by the deliberative layer and, if necessary, to reject them; to recognize recurring patterns in the deliberative subsystem; to develop long-term strategies; and to communicate effectively with others.
Sloman points out that these three layers are hierarchical, but parallel and that they also work parallelly. Like the overall system, these modules possess their own architecture, which can contain further subsystems with their own architecture.
The meta management module is everything else butperfect. This is because it does not have comprehensive access to all internal states and processes, that control over the deliberative subsystem is incomplete, and that the self evaluations can be based on wrong assumptions.
An architecture like the one outlined the so far contains a variety of control states on different levels. Some of them operate on the highest abstraction level, while others are used unconsciously with frequent control decisions.
The following illustration gives an overview over the control states of the system:
Fig. 16: Control states of an intelligent system (Sloman, 1998b, p. 17)
Different control states possess also different underlying mechanisms. Some can be of chemical nature, while others have to do with information structures.
Control states contain dispositions to react to internal or external attractions with internal or external actions. In the context of the overall system, numerous control states can exists simultaneously and interact with one another.
Control states are known in Folk Psychology under numerous names: desires, preferences, beliefs, intentions, moods etc.. By the definition of such states through an architecture, Sloman wants to supply a "rational reconstruction of a number of everyday mental concepts".
Each control state contains, among other things, a structure, a transformation possibility and, if necessary, also a semantic. Sloman illustrates this by the example of a motivator (see below):
Additionally, control states differ in the respect whether they can be changed easily or only with difficulty. Many control states of higher order, so Sloman, can be modified only in small steps and over a longer period. Besides, control states of higher order are more powerful and more influential regarding the overall system than control states of a lower order.
Sloman postulates a process called circulation, by which the control states circulate through the overall system. Useful control states can rise upward in the hierarchy and enlarge their influence; useless control states can disappear from the system nearly completely.
The result of all these processes is a kind of diffusion with which the effects of a strong motivator distribute themselves slowly into countless and long-lived control sub-states, up to the irreversible integration in reflexes and automatic reactions.
A central component of every intelligent system are motivators. Sloman defines them as "mechanisms and representations that tend to produce or modify or select between actions, into the light of beliefs." (Sloman, 1987, p. 4).
Motivators can develop only if goals are present. A goal is a symbolic structure (not necessarily of physical nature) which describes a condition, which is to be achieved, received or to be prevented. While beliefs are defined by the fact that they are representations which adapt by perception and deliberative processes to reality, goals are representations which elicit a behavior in order to adapt reality to the representation.
Motivators are generated by a mechanism which Sloman calls motivator generator or motivator generactivator. Motivators are generated due to external or internal information or produced by other motivators. Sloman defines a motivator structure formally over ten fields: (1) a possible condition which can be true or wrong; (2) a motivational attitude towards this condition; (3) a belief regarding this condition; (4) an importance value; (5) an urgency; (6) an insistence value; (7) one or more plans; (8) a commitment status; (9) management information and (10) a dynamic state like e.g. "plan postponed" or "currently under consideration".
In a later work (Sloman, 1997c), Sloman extended this structure by two further fields: (11) a rationale, if the motivator developed from an explicit thought process as well as (12) an intensity which specifies whether a motivator already worked on gets further preference over other motivators.
The strength of a motivator is determined by four variables:
Motivators compete with one another for attention, the limited resources of the deliberative sub-system. To make this sub-system work, there must be a mechanism which prevents new motivators from getting attention at any time. For this purpose the system posseses a so-called variable threshold attention filter.
The filter specifies a threshold value a motivator must pass in order to get attentional resources. This filter is, as already implied by its name, variable and can be changed, for example by learning. Sloman illustrates this by the example of a novice driver who cannot converse with someone else while driving because he has to concentrate too much on the road. After a certain practice this is, however, possible.
The insistence of a motivator, and thus the crucial variable for the passing of the filter, is a quickly computed heuristic value of importance and urgency of the motivator.
If a motivator has surfaced and thus passed the filter, several management processes are activated. Such a management is necessary because several motivators always pass the filter simultaneously. These processes are adoption-assessment (the decision whether a motivator is accepted or rejected); scheduling (the decision, when a plan is to be executed for this motivator); expansion (developing plans for the motivator) as well as meta-management (the decision whether and when a motivator is to be considered at all by the management).
Sloman's attention filter penetration theory requires a higher degree of complexity than the theory of Oatley and Johnson-Laird. He postulates that not every motivator interrupts the current activity, but only such which exhibit either a high degree of insistence or for which the appropriate attention filters are not set particularly high.
A system that has to survive in an environment which changes continually needs a mechanism with whose assistance it can react without delay to such changes. Such a mechanism is an alarm system.
An alarm system is not only of importance for a reactive, but likewise for a deliberative architecture. For example, the planning ahead can show a threat or a possibility which can be answered immediately with a change of strategy.
Sloman draws a parallel between his alarm system and neurophysiological findings:
The different layers of the system are influenced by the alarm system, but in different ways. At the same time they can also pass informations to the alarm system and thus elicit a global alarm.
For Sloman, emotions are not independent processes, but develop as emergent phenomenon from the interaction of the different subsystems of an intelligent system.
Therefore, no necessity exists for an own "emotion module". A look at psychological emotion theories leads Sloman to the conclusion:
If one, however, views emotions as the result of an accordingly constructed architecture, then, according to Sloman, many misunderstandings can be cleared up. A theory which analyzes emotions in connection with architectural concepts is for him therefore more effective than other approachese:
The different layers of the outlined architecture support also different emotions. The reactive layer is responsible for disgust, sexual arousal, startle and fear of large, fast approaching objects. The deliberative layer is responsible for frustration through failure, relief through danger avoidance, fear of failure or pleasant surprise by a success. The meta-management layer supports shame, degradation, aspects of mourning, pride, annoyance.
Sloman's approach intentionally disregards physiological accompaniments of emotions. For him these are only peripheral phenomena:
Sloman also does not accept the objection that emotions are inseparably connected with bodily expressions. He counters with the argument that these are only "relics of our evolutionary history" which are not essential for emotions. An emotion derives its meaning not from the bodily feelings which accompany it, but from its cognitive content:
He argues in a similar way regarding a number of non-cognitive factors which could play a role with human emotions, for example chemical or hormonal processes. He asks whether the affective states elicited by such non-cognitive mechanisms are really so different from those which are produced by cognitive processes:
How, then, do emotions develop in Sloman's intelligent system? Basically, he differentiates between three classes of emotions which correspond to the three layers of his system. On the one hand, emotions can develop through internal processes within each of these layers; on the other hand by interactions between the layers.
Emotions are accompanied frequently by a state which Sloman calls perturbance. A perturbance is given if the overall system is partially out of control. It arises whenever a rejected, postponed, or simply undesirable motivator emerges repeatedly and thus prevents or makes more difficult the management of more important goals.
Of crucial importance here is the insistence value of a motivator which for Sloman represents a dispositional state. As such a highly insistent motivator can elicit perturbances even then if it has not yet surpassed the filter or is not yet worked on actively.
Perturbances can be occurrent (attempt to attain control over attention) or dispositional (no attempt to attain control over attention).
Perturbant states differ by several dimensions: Duration, internal or external source, semantic content, kind of disruption, effect on attentional processes, frequency of disruption, positive or negative evaluation, development of the state, fading away of the state etc..
Perturbances are, like emotions, emergent effects of mechanisms whose task it is to do something else. They result from the interaction of
For the emergence of perturbances, one thus does not require a separate "perturbance mechanism" in the system; also questions about the function of a perturbant state are not meaningful from this point of view. Perturbances, however, are not to be equated with emotions; they are rather typical accompaniments of states which are generally called emotional.
For Sloman, emotional states are, in principle, nothing else than motivational states caused by motivators.
A further characteristic of emotional states consists in the production of new motivators. If, for example, a first emotional state resulted from a conflict between a belief and a motivator, new motivators can develop which lead to new conflicts within the system.
Sloman and his working group have developed a working computer model named MINDER1 in which his architecture is partly implemented. MINDER1 is a pure software implementation; there is thus no crawling room with real robots. The model is described here very shortly; a detailed description can be found in [Wright and Sloman, 1996].
MINDER1 consists of a kind of virtual crawling room in which a virtual nanny (the minder) has to watch out for a number of virtual babies. These babies are "reactive minibots" which always move around in the crawling room and are threatened by different dangers: they can fall into ditches and be damaged or die; their batteries can run dry, thus they have to get to a recharging station; if the batteries are too much emptied, they die; overpopulation of the crawling room turns some babies into rowdies which damage other babies; damaged babies must be brought into the hospital ward to be repaired; if the damage is too great, the baby dies.
The minder now has different tasks: It must ensure that the babies lose no energy, that they do not fall into a ditch or are threatened by other dangers. For this purpose it can build, for example, fences to enclose the babies therein. It must lead Minibots whose energy level is dangerously low to a recharging station or others away from a ditch as far as possible.
This variety of the tasks ensures that the minder must always produce new motives, evaluate them and act accordingly. The more Minibots enter the crawling room, the less the efficiency of the minder.
The architecture of MINDER1 corresponds to the basic principles described above. It consists of three subsystems which contain themselves a number of further subsystems.
The reactive sub-system contains four modules: Perception, belief maintenance, reactive plan execution, and preattentive motive generation.
The perception subsystem consists of a data base which contains only partial information about the environment of the minder. The system functions within a certain radius around the minder, but can not detect, for example, hidden objects. An update of the data base looks as follows:
time 64 name minibot4 type minibot status alive distance 5.2 x 7.43782 y 12.4632 id 4 charge 73 held false]
This means: Information at time 64 about the minibot named minibot4: It lives, is situated at a distance of 5.2 units from the minder, has the ID 4 and the charge 73 and is not held by another agent.
The belief maintenance subsystem receives its information on the one hand from the informations of the perception subsystem, on the other hand from a belief data base in which, for example, is stored that fences are things with which one can secure a ditch. In order to delete wrong beliefs from the system, every belief is assigned a defeater. If the defeater is evaluated as true, then the respective belief is deleted from the respective data base. An example:
[belief time 20 name minibot8 type minibot status alive distance 17.2196
x 82.2426 y 61.2426 id 8 charge 88 held false
[[belief == name minibot8 == x ?Xb y ?Yb ==]
[WHERE distance(myself.location, Xb,Yb) < sensor_range]
[NOT new_sense_datum == name minibot8 ==]]]]
The defeater in this case means: "IF I possess a belief regarding minibot8 AND I have no new perception data of minibot8 AND I am at a position, in which I should have according to my belief new perception data of minibot8 THEN my belief is wrong."
The subsystem of the reactive plan execution is necessary, so that the minder can react fast to changing external conditions. If it has the plan, for example, to move from one position in the crawling room to another, then this plan should be executed without using too many resources.
To achieve this, MINDER1 uses a method which was developed by Nilsson (1994) and is called teleo-reactive (TR) program formalism. MINDER1 has thirteen of such TR programs which enable it, for example, to look for objects or to manouvre in the room.
In order to use TR programs, the minder first needs to have goals. These are produced by the subsystem for pre-attentive motive generation which consists of a set of generactivators. An example is the generactivator G_low_charge, which searches through the belief database after information about babies with low charge. If it finds such an information, it forms from it a motive and deposits it in the motive data base. An example:
[MOTIVE motive [recharge minibot4] insistence 0.322 status sub]
The status sub denotes that the motive has not yet passed the filter. MINDER1 contains eight generactivators which express its different concerns.
The deliberative sub-system of MINDER1 consists of the modules filter, motive management, and plan excution. All these modules are shallow, thus possess little depth of detail.
The filter threshold in MINDER1 is a real number between 0 and 1. A motivator with the status sub can pass it, if its insistence value is higher than the value of the filter threshold. The status of the motivator then changes from sub to surfacing. A motivator which does not succeed in passing the filter during a time cycle, can be sent back by the generactivator with a newly computed insistence value.
All motivators which have passed the filter are processed by the motive management and receive the status surfaced. The motive management works with the three modules deciding, scheduling, and expanding.
The module "deciding" specifies whether the motivator is to be worked on immediately or later. If it is processed directly, it receives the status active; if it is processed later, it receives the status suspended.
Such a decision is possible mostly only then if the motivator has been submitted to a closer inspection, This is done by the module "expanding". The motivator is expanded in a way that it contains a meta plan. An example:
[MOTIVE motive [save ditch1 minibot5] insistence 0.646361 status active]
is partially expanded to:
[MOTIVE motive [save ditch1 minibot5] insistence 0.646361 status active
plan [[decide] [get_plan]]
The steps decide and get_plan of the meta plan are calls for management processes.
From the module "expanding" the motivator can be executed directly (status: executed), a stored plan can be called up (status: active, meta), or the motivator can be passed on to the module "deciding" (status: active, meta).
In the module "deciding", the active motivator is submitted to a routine which specifies its importance. This importance is determined heuristically from the insistence value; the determination of the urgency of a motivator is not yet implemented in MINDER1. As a result, it can either be rejected (status: suspended, meta) or processed further (status: active, meta).
Finally, the subsystem of plan execution is implemented in a set of TR programs which can call up other TR programs. MINDER1 contains nine action primitives.
The meta-management sub-system of MINDER1 contains two modules: the change of the filter threshold value and the discovery of perturbant states.
The filter threshold value module observes the number of motivators which succeed in passing the filter. If they are more than three, the threshold value is raised until only three or fewer motivators are processed. Motivators with the status suspended which have already been processed, are assigned the status sub again if their insistence value is lower than the threshold value of the filter. Only if less than three motivators are worked upon, the threshold value is lowered again.
The interaction of this module and the generactivators which compute the insistence values of the motivators anew, results in a continuous movement of the motivators from the pre-attentive into the attentive state and back.
The system is, in this form, not yet able to produce perturbances. Sloman speaks in connection with MINDER1 therefore of "proto-perturbances". In order to produce proto-perturbances, Sloman uses a trick which becomes necessary by the lack of complexity of the system. Motivators which refer to damaged minibots (baby), receive per definitionem a high insistence value; the management processes however assign a lower degree of importance to these motivators - thus a deviation from the normal behaviour of the management processes, during which importance is evaluated according to the insistence.
The respective module in the meta-management sub-system was designed in such a way that it calculates the rate of rejection of motivators. If this value passes a certain threshold, then a proto-perturbant state has occurred.
MINDER 1 does indeed show such proto-perturbances. However, the sub-system cannot deal further with this information; for this, the entire system is not yet developed enough.
Sloman's theoretical approach is certainly one of the most interesting regarding the development of emotional computers. It is less his specific interest in emotions but rather his stressing of the architecture which opens up new perspectives.
Sloman follows through theoretically most consistently what others had speculated about as well: that there is no fundamental difference between emotion and cognition. Both are aspects of control structures of an autonomous system.
A detailed view of Sloman's work from 1981 to 1998, however, shows a number of ambiguities. For example, the differentiation between the terms goal, motive, and motivator is not clear, because they are used by him quite interchangeably.
Also it does not become clear what function perturbances have exactly with the emergence of emotions and how they are connected with the global alarm system postulated by him. It is interesting that in his earlier work this alarm system is scarcely mentioned, but mainly perturbances; in his later work one finds nearly the opposite.
The proof which Sloman wanted to deliver with MINDER1 is, in its present form, not convincing. Neither do perturbances develop from the interaction of the elements of the system (the programmers had to help a lot to produce even proto-perturbances), nor can one draw from it far-reaching conclusions about human emotions.
It is nevertheless the theoretical depth and width of Sloman's work which can lend new impulses to the study of the emotions. His combination of design-oriented approach, theory of evolution and discussion of virtual and physical machines is deeper than all other approaches for the construction of autonomous agents.