Butanalogy can also operate in mutual alignment1 analogies to reveal commonalities thatwere previously not obvious in either analog.

Projecting information from a well-understood domain can lend structure to an unfamiliar domain, as in:The mitochondria are the power supply for a cell.

Analogy is often thought of chiefly as a way to transfer knowledge from one situationto another, and indeed, it often serves that function.

Relational categories have been the focus of much recent research (Asmuth &Gentner, 2017; Gentner, 2005; Gentner & Kurtz, 2005; Goldwater & Markman, 2011;Markman & Stilwell, 2001; Ross & Murphy, 1999), in part because of their importantrole in conceptual learning and education (Goldwater & Schalk, 2016).

For example, carnivore andherbivore are abstract relational categories, while canine and feline are abstract entity cat-egories.

Relational cate-gories are categories for which the basis for membership is participation in a commonrelational structure; thus, they differ from the more studied entity categories, such as tulipand spoon, whose members share many intrinsic properties.

Our main focus is on relational abstractions, includingprinciples, rules, and schemas, as well as abstract relational categories.

For example, causal system is more abstract than posi-tive feedback system, which in turn is more abstract than the specific positive feedbacksystem by which the melting of polar ice causes lower reflectance of the sun’s heat, lead-ing in turn to more rapid melting.

Wetake the process of abstraction to be one of decreasing the specificity (and therebyincreasing the scope) of a concept.

Many such abstractions are expressed as rela-tional categories—categories like evidence, counterfactual, and proportion, and on a moremundane level, bargain, ally, and rescue.

Theassertions that make up abstract knowledge are variously referred to as schemas, rules,abstractions, principles, or overhypotheses

Abstract structured knowledge is a key feature of higher order cognition (Gentner &Medina, 1998; Hummel, 2011; Markman, 1999; Tenenbaum & Griffiths, 2001).

Long ago my forefathers came across the sea. Far they came, in white ships tall as trees and on the land they built them wagons and covered them with sails of their ships. Far they travelled and spread their campfire ashes over this vast barbaric land. But now their children are tired, we want to build

Engineering refers to the use of technical principles, such as mathematics, science, and technical know-how, to realize a design that best meets a given set of expectations, which are typically captured in a requirements specification.

Designing is the process of arriving at a plan, specification, prototype, system, or service—a design. In HCI, this often means designing a user interface and relevant parts of the underlying interactive system.

HCI focuses on people who use an interactive system or are affected by its use. This focus is often called being user-centered or human-centered to contrast it with a focus on the technology itself [423, 604].

Finally, interaction often involves co-adaptation between people and computers [646], meaning that both the user and the system learn and adapt to each other during interactions.

Interaction is, in other words, not a property of the system design or the user but something that emerges when they influence each other.

The development of technology for interactive computing systems has been an important driver behind the widespread adoption of computing we have witnessed in the last 50 years.

In HCI, evaluation refers to the application of some systematic methodology to attribute human-related values to an artifact, prototype, system, or process. Examples of such attributes include performance, experience, safety, and ethical aspects, such as the avoidance of bias or harm.

Programmability lends computers their power as tools. Computer programs can decompose complex activities into sequences of much simpler operations.

A special part of a computing system is the user interface. It is the part that the user can see and utilize to control the computer. Through the user interface, users can provide input and instructions to a computer and receive feedback from it. In short, the user interface enables interaction with a computer.

Beforethen, the terms man–computer interaction and man–machine interaction had been in use sincethe early 1960s [e.g., 476, 588].

Interaction is a concept that is fundamental in HCI and specific to this field [357]. Intuitively, it refers to the reciprocal influence between people and an interactive system that takes place through the user interface.

At the core of this revolution was the ability to flexibly define and execute computer programs—sequences of logical operations executed by a computer.

They enhance our physicalabilities and are central to many intellectual activities, such as writing, mathematics, and account-ing.

From fishing nets to drilling machines, tools are vital to human ability.

A key technical construct in HCI is the user interface. It refers to the parts of an interactive system that the user comes into contact with or that in other ways shape the user's perception of the system.

Second, the computer is among the most complex tools humans have devised.

In HCI, evaluation refers to the application of some systematic methodology to attribute human-related values to an artifact, prototype, system, or process.

The development of technology for interactive computing systems has been an important driverbehind the widespread adoption of computing we have witnessed in the last 50 years.

A special part of a computing system is the user interface. It is the part that the user can see and utilize to control the computer.

Programmability lends computers their power as tools.

It is an egocentric fallacy to assume that others are like us—to attempt to explain other people by reference to one's own experience.

Set the `expectedInputs` and `expectedOutputs` modalities and languages when creating your session

The Prompt API uses the Gemini Nano model in Chrome.

The agent chooses services to use from this catalog based on what the user has asked them to do and the user’s preferences — but the user needs no prior knowledge of what services are offered by which providers, and does not need to provide any input.

When the agent chooses a service and provisions it (ex: `stripe projects add cloudflare/registrar:domain`), it provisions the resource within a Cloudflare account.

The protocol accounts for this in two ways. When an agent provisions a paid service, Stripe includes a payment token in the request to the Provider (Cloudflare). Raw payment details like credit card numbers aren’t ever shared with the agent.

These build on prior art and existing standards like OAuth, OIDC and payment tokenization —but are used together to remove many steps that might otherwise require a human in the loop.

We built AI into our editor's foundation instead of bolting it on top.

The rewards were applied only in the Nerdy condition, but reinforcement learning does not guarantee that learned behaviors stay neatly scoped to the condition that produced them.

We realized we were optimizing the wrong thing. We were orienting our system around coding sessions and merged PRs, when PRs and sessions are really a means to an end.

Erdős also came up with the Erdős sum, a 'score' you can calculate for any primitive set. He showed that the sum had a maximum possible value—and conjectured that this value must hold only for the set of all prime numbers.

The question Price solved—or prompted ChatGPT to solve—concerns special sets of whole numbers, where no number in the set can be evenly divided by any other.

Information scent refers to a user's intuition that a cue in the interface represents the information needed. It is an estimation of relevance based on a proximal cue.

Information foraging refers to information-seeking activities such as navigating, exploring, comparing, searching, or manipulating information contents in an information space.

A payoff refers to the benefits that are left after the costs have been subtracted.

The space of all states, actions, and rewards is called the environment.

Costs are negative rewards that the user incurs for transitioning between states or for being in states that are not good for them.

Visual statistical learning is a research topic in perception that studies how the statistical distribution of our environments affects the deployment of gaze.

bounded rationality states that we are only rational to the extent allowed by the involved constraints, or bounds.

An agent is an actor with the ability to choose actions in pursuit of some goal or reward.

To state that a user's choice is rational means that it is selected with the expectation that it yields the highest utility out of the available options.

Utility refers to the agent's consideration of positive and negative rewards when deciding how to act.

The term satisficing is used to describe how users tend to behave when facing a complex decision-making problem. It refers to settling on a satisfactory but not optimal solution in the normative sense.

The concept of rationality has its roots in economics, where it was developed to study how peo-ple should act in economic decision-making. In such settings, the idea is that people reach theirgoal, such as maximizing their return, by maximizing utility.

This chapter introduces the notion of rationality. This allows us to provide explanations forinteractive behavior due to users attempting to make the most out of the choices available to them.

That’s up 20x in six weeks. This idea, called tokenmaxxing, is the deliberate practice of maximizing token consumption.

A wire becomes a transmission line. A bend becomes a reflector. Two parallel traces become coupled antennas. The geometry is the circuit.

We find internal representations of emotion concepts, which encode the broad concept of a particular emotion and generalize across contexts and behaviors it might be linked to.

The difference between analysis automation—inference—anddecision automation is that in the latter the system must make implicit or explicit assump-tions about costs and values inherent in all decisions.

Examples of decisionautomation include route planning and adaptation, such as to avoid bad weather, and systems pro-viding medical diagnosis support.

Decision automation means deciding and selecting appropriate actions among alternatives.This type of automation corresponds to the third human information processing state, decision-making, which the machine is either augmenting or replacing altogether.

An example of low-level automation is the extrapolation or prediction of data over time,such as a system predicting a trend for the output of an industrial plant based on historical sensordata. An example of moderate- to high-level automation is a system integrating multiple sources orinput variables. This could be a display with emergent perceptual features, such as an optical see-through display with a landing strip intended to assist a pilot in landing an aircraft. An exampleof high-level automation is a context-dependent summary of data.

Analysis automation refers to the automation of information analysis and involves inferentialprocesses. It corresponds to the second human information processing state: perception/workingmemory.

An example of low-level automation is assistance in sensor adjustment, such as a system mechanically moving a radarsensor to lock on a detected target. An example of moderate automation is a system organizinginformation according to criteria such as a priority list or highlighting information based on staticor dynamic criteria. This could be, for example, a display highlighting the rate of change in somevariable of interest. This could be indicated by increasing the intensity of some pixels more rapidlythan others in the display.

Acquisition automation corresponds to the first human information processing stage, sensoryprocessing, and it is realized by the system sensing and registering input data.

All three strategies have a common deficiency in that they may not always be able to adhereto the principle of human-centered automation, whereby a human has final control. This is alsocalled the authority problem.

The third allocation strategy is to allocate each function in a way that maximizes economicefficiency.

The second allocation strategy is assigning each function to the most capable agent, which canbe either a human or a machine.

Therefore, this automation strategy definesthe roles and responsibilities of users in terms of automation instead of the other way around.

First, byautomating everything that can be automated, the user is left with functions that, by definition, thedesigners find hard, expensive, or difficult to automate.

The first strategy is called maximum automation. Here, each task that can be automated is allo-cated to a machine.

The aim is to increase efficiency, reduce costs, or both.

Which system functions should beautomated, and to what extent should they be automated?

Task allocation is a central challenge in HCI and automation.

Four levels of shared control can be distinguished [1]: strategic (e.g.,setting a destination), tactical (e.g., doing a specific maneuver like merging into a lane), oper-ational (e.g., maintaining a certain distance from another car), and execution (lowest-level ofcontrolling locomotion, steering, and so on).

Control does not need to be either/or like in many semi-autonomous vehicles.

When two agents sharing control have asymmetric capa-bilities, both loose and tight rein control should be available.

First, control can be shared via an extensionthat allows a machine to amplify human ability.

When riding a horse, the rider communicates high-level information (e.g., the goal) to thehorse but must be ready to guide the horse at a lower level. When the horse knows what to do,for example, if the route is familiar, the rider may not need to engage in low-level control. Thisform of control, called loose rein control, is possible if the horse knows what the rider wants.

First, communication is vital for sharingcontrol, and this can happen at different levels; second, both agents must have internal modelsof each other to understand what those communicative acts mean.

Shared control is about carrying out a task together with a competent partner [1, p. 511]:“In shared control, human(s) and robot(s) are interacting congruently in a perception–actioncycle to perform a dynamic task that either the human or the robot could execute individuallyunder ideal circumstances.”

The question of shared control is timely; semi-autonomous vehicles are only partiallyautonomous. They need the human to assist them and, therefore, some way of handing controlover to the human driver. They also need to have guidance from the driver, for example, onthe choice of route.

Independently of the chosen strategy, some tasks are to be done by the interactive system andsome by the user. The allocation of such tasks is called functional allocation.

An example of such control sharing is powersteering in a car: The car provides additional work to allow the driver to turn the wheels withless effort. An HCI example is mouse acceleration, which allows a user to move the cursor on thescreen farther than the physical movement of the mouse.

Second, control can be shared via relief, which means that the overall burden on the humanis reduced by the machine. An example is automatic shift transmission, which relieves the driverof the task of changing gears in a car. An HCI example is text entry using autocomplete, whichprevents the user from correcting typing mistakes as they type.

Third, control can be shared via partitioning. In this case, a task is decomposed into parts thatcan be addressed by humans and machines separately. An example of such control sharing is semi-automatic parallel parking, which provides the driver with some braking ability while the machinecontrols the speed and steering of the car. An HCI example is automatic spell checking, where thesystem detects and highlights incorrectly spelled words but does not change them. Instead, theuser has to take an explicit corrective action, such as selecting a misspelled word and choosing analternative.

In a task-switching situation,the user must activate resources for the second task and inhibit resources for the first task. If theuser fails to do so efficiently, performance is reduced, sometimes dramatically.

Successful time-sharing depends on the strategy and difficulty of the task in terms of tempo-ral constraints—how many tasks are processed in a given interval—and task complexity—thequantity of information that needs to be processed for a given task.

The first is called thesingle channel theory, which posits that there is limited capacity in the human information pro-cessing system in a time-sharing scenario. When the channel capacity is exceeded, multiple taskstransition from parallel processing to serial processing.

If both tasks demandcontrolled processing, then the strategy in processing is split into two mechanisms: facilitationand inhibition.

The third theory is information processing analysis theory. Ifat least one task can be carried out automatically, the other task can be carried out with little orno impact on performance (at an appropriate time–error trade-off point).

The second theoryis the multiple resources model, which states that resource limitation concerns the entire systemrather than a channel (Chapter 5).

Second, tasks canbe shared in terms of control, which means that some control over the tasks is assigned to anotheragent, such as another user or a machine.

Ingeneral, perfect time-sharing with no degradation in performance occurs only for tasks that areautomatic, such as speaking while walking. Such tasks can reach automaticity.

Ingeneral, perfect time-sharing with no degradation in performance occurs only for tasks that areautomatic, such as speaking while walking.

First, tasks can be time-shared, which implies that the user performs multiple tasks.

For the sharing of tasks tooccur, there have to be at least two tasks or subtasks, which can then be shared in two differ-ent ways.

A central concern in the design of automation and AI is task sharing.

In all cases, getting the right balance betweenautomation and human control is crucial.

Throughout the history of automation, a central challenge in HCI has been posed by the com-plexity of autonomous systems. Users must be able to understand and control them. They need tofind and integrate information from dynamic and different sources. They need to understand howto delegate tasks, supervise their execution, and intervene if needed.

Automation refers to technology that assists users by performing a task or a subtask on their behalf.

Direct manipulation does not anthropomorphize user interfaces or deskill users.

Direct manipulation allows the user to predict what will happen, explore the system, andfeel in charge.

Software agentsmay allow users to achieve their goals in complex environments with limited expertise.

In HCI, automation poses the foun-dational problem of task allocation: Who does which task or subtask—the system or the user?Moreover, how much should we automate?

Cai et al. [117] interviewed 21 pathologists who used a deep neural network to aid in thediagnosis of prostate cancer. The interviews showed that pathologists needed to learn moreabout the network’s strengths and limitations to use it effectively. They also wanted to knowthe design objective of the network and the kind of data on which it was trained.

Roy et al. [715] explored what happens when users doing a task with a simulated crane needto choose between further using automation and manually continuing the task. The authorsshowed that the decision of whether to use automation is affected by the users’ perceptionof its accuracy as well as how easy it is to do the task themselves.

In other scenarios, the agent may be continuously monitoring and analyzing theuser’s actions and proactively acting to assist the user.

nteraction between usersand agents opens up new ways of thinking about interaction.

Automation technology is capable of doing things on its own. Wecall such technology an agent, also sometimes called a software agent, intelligent agent, conversa-tional agent, personal digital assistant, or intelligent interactive system.

Wecall such technology an agent, also sometimes called a software agent, intelligent agent, conversa-tional agent, personal digital assistant, or intelligent interactive system.

Automation technology is capable of doing things on its own.

An exciting and distinctive aspect of this type of interaction is that users interact with some-thing that may have agency.

They may need to hand over control to a human driver in parts they cannothandle.

Semi-autonomous vehicles can drive parts of aroute autonomously.

They typi-cally do so by using some kind of AI that takes over part of a task from the user.

Cognitive integration means that we internalize the operation of the tool. We not only act but also start thinking as defined by the unique constraints and mechanisms of the tool.

accessibility concerns the match between a user's abilities and the system's required abilities. As such, it differs from usability (which is about the relationship between users, tools, and tasks) and utility (which is about whether a tool may be used to complete a task).

TTF refers to the ability of technology to support a task. The capabilities of the technology should match the demands of the task and the skills of the individual; in this case, the fit is perfect.

TAM posits that the intention to adopt a particular technology is driven by two kinds of perceptions: (1) how easy it is to use a system and (2) how useful it will be to use it. Furthermore, the perceived ease of use affects the perceived usefulness: If technology is hard to use, it is less useful.

The second dimension, social acceptability, concerns whether interactions map well to the social norms and roles in the settings where they occur.

Acceptability has two main dimensions. The first dimension, practical acceptability, includes costs, the reliability of the interactive system, and its compatibility with other systems. The perceptions of utility and usability may also influence the judgment of practical acceptability.

usability is multidimensional. This means that in most settings, a valid characterization of usability will need to employ several dimensions and measures.

usability is measurable, that is, it is possible to quantify usability based on users' behaviors or opinions.

usability is relational; it arises as an interplay between people, tasks (problems), and interactive systems (tools)

The utility of an interactive system concerns its match with the tasks of users. If the match is good, the tool has high utility; if the tasks that users want to do are not supported by the tool, the tool has low utility.

Usability concerns how easily computer-based tools may be operated by users trying to accomplish a task. Usability differs from utility. Usability concerns whether users can use the product in a way that makes it possible to realize its utility; utility is about whether the goal is important to the user.

Robustness refers to the communication partners' ability to achieve shared understanding even in light of misunderstandings and other unanticipated troubles.

Communication repair refers to the "work of restoring shared understanding" when conversational partners misunderstand each other.

Mixed-initiative interaction is the idea of organizing interaction in dialogue where both the computer and the human can take initiative.

Gulf of evaluation: This gulf refers to knowing how a perceived change in the computer has moved it closer to the intended goal state.

Gulf of execution: This gulf is about knowing what to do to bring about a desired state change in the computer.

Code-switching refers to a switch in language to match the capabilities of the communication partner.

An FSM is a model of discrete computation applicable to dialogues. In computer science, an FSM is a special case of a Turing machine that reads but does not write on the tape.

A mode refers to the variation in the interpretation of a user's input according to an internal state. In a modeless dialogue, all inputs are possible in all states and their interpretation is always the same.

The key idea in the dialogue view of interaction is the organization of communication as a series of turns. Dialogue evolves through communication turns between two or more partners. In one turn, an appropriate communication act is made by one partner based on the communication context. The act aims to get the other partner to do or understand something. This understanding then forms the context within which the other partner takes their turn.

Interaction may be viewed as a dialogue, that is, a conversation that occurs between two partners in a context for some purpose.

For example, in control-theoretical analysis, we see the interaction as dynamically changing states.

For example, in interaction-as-rationality, courses of action emerge as a joint function of the user's goals and capabilities and the properties of the environment.

For example, statistical models, such as Fitts' law, describe a relationship that is considered a statistical determination.

For example, interaction-as-tool-use focuses on this idea (Chapter 19). Tools change people and their activities; in turn, this changes the tools.

An interactive AI system can possess an objective function it pursues when acting, for example, when correcting a character it flags as a typo.

In the case of particle physics, the propositions that make up a theory may concern the nature and behavior of particles.

HCI theories contain statements that link humans and technology and possibly some outcomes (e.g., poor usability, high user experience).

Suchtechniques provide different levels of automation.

tool use allows us to assess text entry in terms of its utility—how well it supports users in thatpursuit—and its usability—how easy it is to enter text and learn how to enter text.

For example, word predictionusage may not be beneficial if the word suggestions are poorly suited to the individual’s language orif the user can type words faster than the time required to scan and select among word suggestions.

The effect of such automation critically dependson how well its design appreciates the actor’s unique capabilities.

Text entry can also be seen as a task where different subtasks are shared between the human and the computer (Chapter 20).

One example is autocorrect, which automatically corrects typing errors while the user is typing. Another example is the use of word predictions, which allow the user to select a word from a set of word suggestions instead of typing out the word in full.

For example, text entry methods such as eye typing are designed to allow nonspeaking users with motor disabilities to enter text using their eye movements only.

Persons with varying capabilities and backgrounds need toeffectively enter text, or these systems fail as tools.

Tool use alsotriggers the concept of accessibility.

Thinking about text entry as

Text entry is also a good example of tool use (Chapter 19). A text entry method is a tool that allows the user to communicate with someone or something, typically other people or a service, using asynchronous text messages and longer documents.

Text entry can also be carried out via spoken dialogue.

When you enter a letter (a turn), a small pop-up appearsfor you to confirm your press (a turn). A word prediction list may show possible completions of theword. At every press, these are updated, over time forming a graphical dialogue between the userand the system.

The essence of dialogue is that communi-cation is organized in sequences of turns.

Moreover, control theory offers a broad view of interaction; itis not limited to input.

Viewing text entry as a control problem can providea more in-depth understanding of the human perceptual and motor control aspects of text entry,such as a user’s ability to touch individual letter keys on a touchscreen keyboard or a user’s abilityto perform a touchscreen gesture.

Interaction is understood as actions and reactions that minimize the discrepancybetween the present state and the goal state.

For example, the information rate ofa text entry method can be quantified using the concept of throughput from information theory.

Informationtheory provides a rigorous formalism to understand and quantify interaction via the concept ofpassing messages through a noisy channel.

This con-cept, communication of information, is rooted in information theory (Chapter 17).

Accessibility is an extension of usability to ensure that as many people as possible find a tool easy to operate, regardless of individual capabilities.

Usability refers to the ease with which a tool is operated.

Utility captures how well a tool supports users in achieving their goals.

A model is a formally expressed set of propositions that follows some axiomatic system, such as algebra, logic, or a programming language.

The explanatory power of a theory refers to the empirical accuracy and coverage of the explanations offered by the theory.

A latent factor is something that affects observations about interaction without being directly observable.

Finally, in quantitative determination, interaction is described as a continuous unfolding of states.

In structural determination, the end results of the interaction are jointly determined by multiple causes that make up the whole.

In statistical determination, there is a stochastic relationship between the two entities.

In mechanical determination, an antecedent determines a consequent.

Propositions characterize entities and link them to other entities, some of which are conceptual.

To conclude, Bunge’stypology is useful in understanding the landscape of HCI theories.

An interactive AI system can possess an objective function it pur-sues when acting, for example, when correcting a character it flags as a typo.

However, computers can also have goals or at least algorithmic objectives. This adds another levelto teleological determination.

All theories of interaction discussed in this part assume intentionality: People have goalsthat have a (mutually) causal role in interaction.

In teleological determination, goals or purposes determine interaction in some way.

One type of determination is shared by all theories of interaction in HCI: teleological determina-tion.

For example, in control-theoretical analysis, we see the interaction asdynamically changing states.

Each state dynamically leads to a new state depend-ing on the forces involved.

Finally, in quantitative determination, interaction isdescribed as a continuous unfolding of states.

For example, ininteraction-as-rationality, courses of action emerge as a joint function of the user’s goals and capa-bilities and the properties of the environment.

In structural determination, the end results of theinteraction are jointly determined by multiple causes that make up the whole.

Here, the statistical model (regression) links the time it takes for a user toelicit a response to the design of the task environment (distance and width of buttons).

For example, statistical models, such as Fitts’ law, describe a relationship that is considered a sta-tistical determination.

In statistical determination, there is a stochastic relationship between the two entities.

The pressing of a button would count as a mechanistic determination.

In mechanical determination, an antecedentdetermines a consequent.

Bunge analyzed theories in scientific fields, from biology to physics, anddeveloped a rich typology of mutual determination.

What happens in interaction is mutually determined by the human and the computer. In other words, what happens in interaction cannot be attributed solely to the human or the computer—the two must be considered together.

For example, they can talk about information, difficulty, working memory, and so on.

Propositionscharacterize entities and link them to other entities, some of which are conceptual.

HCI theories contain statements that link humans andtechnology and possibly some outcomes (e.g., poor usability, high user experience).

In the case of particle physics, the propositions that make up a theory may con-cern the nature and behavior of particles.