Inference Theory: Deductive, Inductive, Abductive And Human Inference

Types of Inference

Question:

Discuss about the Human Inference Theory.

Inferences describe the stages in reasoning that moves from the premises to conclusions. The inference have been divided by Charles Sanders Peirce into 3 kinds of induction, deduction and abduction. The deduction inference refers to deriving logical conclusions from the assumed or known premises to be true with valid inference laws being studied logically. Induction inference is derived from a given premises to the universal conclusions whereas abduction inference is drawn from the best explication. The human inference or how the human arrive at conclusions is studied conventionally within the cognitive psychology field; artificial intelligence scholars establish automated inference systems that facilitate the emulation of human inference.

The statistical inference utilizes the mathematics to derive conclusions in face of uncertainty thereby generalizing the deterministic reasoning without uncertainty as a unique instance. The inductive reasoning describes the process by which a conclusion becomes inferred from manifold observations. This conclusion could either be incorrect or correct within a given degree of accuracy or correct in specific context. Conclusion inferred from manifold observations could be tested by extra observations. Inference is conventionally described as that conclusion arrived at based on reasoning and evidence or the process of arriving at such evidence/reason-based conclusion.  

Human Inference Theory (HIT) encompass a tangle of challenging as well as interrelated issues which entail contributions from diverse disciplines. Inferences are phases in reasoning which assist a person to get to a deduction from the premises. According to correspondent inference theory, the judgment of personality of a human being is stated to correspond to some particular behavior. In the correspondent inference theory, the 2 principles remained vital to infer a character of a person positively correlates with her respective behavior. For example, an action is what many people shall be anticipated to behave in a context and the response recommended traits are never obvious to reasonably draw correspondent inferences. Traditionally, rational conclusions were studied in the discipline of cognitive psychology. Here, the scholars in artificial intelligence (AI) developed a system in automated inference to imitate human reasoning (Goodman & Frank, 2016).

The Greek philosophers argued that any given argument’s validity hinges on deduction form and the phrase valid doesn’t imply that the premises or conclusion remain true. Because statistics remains the study of variable data, their respective measures reflect how the can be modelled effectively. It remains valuable to know that instruments whose scale remain imperfect and, therefore, shall make the outcome vary can measure such a data. Primarily the variability could be triggered by the seasonality of environmental variables which shall occur in the course of the day. All such variability shall culminate in the utilization of probability to make an effective conclusion whereby there is uncertainty is generalizes the deterministic reasoning. The categorical and quantitative data is used in deductions that are subjected to random variables. The likelihood where the researcher is able to estimate utilizing the confidence or utilize the Bayesian probability assists in making the right probability distribution hence writing an effective hypothesis and ultimately hit a conclusion.

Statistical Inference

Human is divided into 3 classes according to Charles Sanders Pierce. The 3 categories encompass deductive, abduction and inductive. Deductive inference are utilized in driving thoughtful conclusions from the underlying assumptions which are known to be true. Moreover, the deductive reasoning can as be divided into 3 schools of thought. The 1^st class is hinged on the performance on factual knowledge. The 2^nd is category is concerned with the system of formal assumptions of reasoning to those in the logical calculus.  The 3^rd category is anchored on mental models of thinking to the ones of in a logical calculus (Heit, 2015).  The deductive inference hits a conclusion in a reductive way whereby it applies general rules which put together the whole domain of enclosure. It narrows this range that is under consideration till the need conclusion is exited. Inductive reasoning is a process whereby several assumptions held to be true are merged together to acquire a given conclusion. The inference is primarily utilized in the applications which depend on predictions behavior/forecasting (Kaplan, 2013).   

Inductive reasoning unlike other logic, is an argument that view premises as proving a substantial evidence that lead to a true conclusion. Abduction is a logical reasoning that engages from what is under observation to a theory meant to offer an account of the observation which ideally avails the simplest explication. As opposed to other reasoning, the premises don’t ideally give an assurance to the conclusion, and, therefore it is able to be understood as an inference to best result. As such, it makes the abduction inference equivalent to the logical fallacy which affirms the subsequent because of manifold conclusion in observation. 

Integrating inference in the construction of artificial intelligence systems occurs in two stages. In the first one, algorithms for fixed functions like mathematical calculations and calendars are written and added to the system (Hosch, 2013). The second one involves writing adaptive algorithms that mimic human inference. People continually adapt to new situations as they encounter them thus infinitely extending their knowledge base. Most computer systems have to be updated with new information and plug-ins to perform functions that were not in the initial system algorithm. Computer scientists are seeking to invent systems that can autonomously extend their knowledge base. Currently, computer systems help in executing numerous complex activities but cannot be used as substitutes for humans because they operate based on whatever information is in their system. They are incapable of processing any information that is not in their knowledge base (Copeland, 2017).

Human Inference Theory

The application of the HIT can best be understood on the basis of the three classes according to Charles Sanders Pierce including the deductive, abduction and inductive inference. An inference is an idea usually derived from the process of reasoning. It involves using empirical data, observation or past experiences to arrive at a logical conclusion about something. Human Inference is a natural part of observation, and it is as old as history (Bradford, 2015).

The human inference theory occurs within the field of cognitive psychology, and it refers to how people draw conclusions while interacting with various stimuli in their environment. According to this model, humans do not spontaneously construct logical proofs when they reason but rather, they rely on rapid intuition making their conclusions prone to systematic errors (Heit, 2015). For this reason, it has become impossible to apply logical formalism in determining and recovering assertions in human inference. However, people can use cognition to correct these errors through deliberation.

Human inference also entails various tasks, some of which can be carried out at the same time. For instance, they can generate conclusions from a set of assertions, consult background knowledge to explain inconsistencies and make probable outcomes for unique events. Humans utilize the use of semantic labels (Heit, 2015). They make predictions based on a categorization of what is immediately recognizable due to a human need for cognitive economy; a term that refers to how people categorize the world to provide a maximally efficient way of representing information about frequently encountered objects (Bradford, 2015).

Inductive reasoning or inference is that reasoning from a particular instance or case and drawing a general rule. It derives inferences from the observation to reach generalization. In so doing, the Inductive Inference recognizes that conclusion could not be sure. The Inductive inference can be used in UAE’s Universities to Attract Students based on four phases: observation; analysis; inference; and confirmation. Under observation, the observation will help the students to gather facts without bias and then undertake the classification of facts, identification of trends or patterns of regularity under the anlysis. Under inference, the students can effective infer generalization from the identified patterns or trends regarding the correlation between facts (Tsuda, 2015).

It is also beneficial to students in arguments. For example, the students will be able to derive a general rule in the conventional area and the move a notch higher to apply such a rule in the field where he wants the person to behave. Also, it will be helpful during the arguments because it will give the students a lot of details and subsequently explicate what it all means in the specific field. Inductive arguments is also helpful to students because it will help them talk about the benefits of the portions and only get to the general benefits later. Inductive inference is helpful in argument as it enables the students to take what has occurred and provide plausible explication for why it has ensued. It gives the learners an opportunity to either use part-to-whole; extrapolation or predictions in arguments (He, 2016).

Application of Inference in AI Systems

Early proponents of induction inference like Francis Bacon have explained the importance of helping the students understand nature in an unbiased manner via the use of Inductive Inference because it derives laws from neutral observation. The induction utilizes evidence as opposed to logic where it states, “all these are true, hence that should be as well be true”. This can lead to uncertain and probabilistic conclusion than the more limited and certain deductive reasoning. The inductive arguments are, therefore, always open to queries as, by definition, the conclusion remains bigger bag as opposed to evidence on which it is founded. Such a breadth permits it to be utilized where the deductive inference fails like in the case of invention and prediction.

In argument, beginning with the comprehensive anchors one’s persuasion in reality, utilizing immediate sensory data of what is seeable and touchable then proceeding to the big picture of ideas, principles and general rules. Beginning from a small and building up to big is able to be less threatening than beginning with big stuff, that can make inductive arguments more convincing  as students might understand the process better as opposed to a more clinical deduction. The inductive inference also attracts scientist since it enables the creation of laws through the observation of a range of phenomena, finding similarities and deriving a law that explain all things. Inductive arguments are thus made to be more valid as well as probable by adding evidence, albeit if such an evidence is selectively chosen, it could falsely hide the opposing evidence. Inductive reason needs both trust and illustration of integrity more than deductive reasoning.  

This is the basic form of valid reasoning. It can as well be called deduction. It begins with a general statement or hypothesis, and then proceed to examine the possibilities to arrive at specific, logical conclusion. The scientific method utilizes deductions to test hypothesis and theories. And since learning most involve hypothesis testing, students will be highly attracted to use deductive inference. Here, the students will be able to hold a theory and based on such a theory, make credible predictions of its consequences. This implies that the students will be able to predict what the observations need to be in case the theory were correct. The students go from a general-theory-to the specific-observation.

Inferences is a basically a set of strategies that help the students develop the essential thinking skills to succeed in tasks. The teachers use inference strategies to appeal to students. All inference strategies work in similar manner: A teacher presents learners with a puzzling query, a discrepant incident, incomplete data, or even an exciting problem to solve. Learners are then anticipated to utilize their reasoning powers to develop hypotheses. The students then test and refine the developed hypotheses. This skill has a broad array significant impact on the achievement of students in UAE’s universities. 

Inductive Reasoning

The students can use different process when drawing their conclusions including inductive learning, mystery, main idea and investigation as inference strategies. Inductive learning helps learners to draw inferences by grouping the data, labelling data groups with descriptive titles, and utilizing groups to produce and test hypotheses. Mystery will present the students with a puzzling question or context and has learners examine clues which help them explicate the mystery (Whitney, 2002). Main idea teaches the learners how to use inferential thinking when constructing main ideas which are not stated explicitly. Investigation challenges the students to utilize a range of problem-solving approaches that need inference. Together, these four inference strategies deepen the interaction of the students with the content and develop their inference skills. They improve students’ thinking and classroom discussion.

Conclusion

Human beings reason subjectively while using their life experiences as a reference (Whitney, 2002). Supposing that two people encounter the same experiences, the meanings each derives from them are unique as our schemas. Scripts are not universal because people react differently to different stimuli. However, they are culturally crafted. For example, a script for eating in the English culture involves placing a napkin on the neck or lap and using a fork and knife. In Chinese culture, food is eaten using chopsticks and so acquiring a script requires learning how to use chopsticks. A person aiming to either understand how people reason as in cognitive psychologists or how to best mimic human reasoning in machines as in computer scientist has to strive to understand how people construct the references they use for inference.

References

Bradford, A. (2015). Deductive Reasoning vs. Inductive Reasoning. Retrieved from https://www.livescience.com/21569-deduction-vs-induction.html

Copeland, B. J. (2017). Artificial Intelligence | Definition, Examples, and Applications | Britannica.com. Retrieved from https://www.britannica.com/technology/artificial-intelligence

Gabbay, D., Hartmann, S., & Woods, J. (2010). Inductive logic (1st ed.). Oxford: North-Holland.

Goodman, N. D., & Frank, M. C. (2016). Pragmatic language interpretation as probabilistic inference. Trends in cognitive sciences, 20(11), 818-829.

Hacking, I. (2001). An introduction to probability and inductive logic (1st ed.). Cambridge, U.K.: Cambridge University Press.

He, X. (2016, February). Understanding Diffusion Processes: Inference and Theory. In Proceedings of the Ninth ACM International Conference on Web Search and Data Mining (pp. 707-707). ACM.

Heit, E. (2015). Brain imaging, forward inference, and theories of reasoning. Frontiers in Human Neuroscience, 8, 1056. Retrieved from https://journal.frontiersin.org/article/10.3389/fnhum.2014.01056

Hosch, W. L. (2013). Genetic Algorithm | Computer Science | Britannica.com. Retrieved from https://www.britannica.com/technology/genetic-algorithm

Kaplan, M. F. (Ed.). (2013). Human judgement and decision processes. Academic Press.

MacKay, D. (2003). Information theory, inference, and learning algorithms (1st ed.). Cambridge, UK: Cambridge University Press.

Rugg, G. (2013). Schema theory, scripts, and mental templates: An introduction | hyde and rugg [Web log post]. Retrieved from https://hydeandrugg.wordpress.com/2013/08/30/schema-theory-scripts-and-mental-templates-an-introduction/

Tsuda, I. (2015). Logic Dynamics for Deductive Inference—Its Stability and Neural Basis. In Chaos, Information Processing and Paradoxical Games: The Legacy of John S Nicolis (pp. 355-373).

Whitney, P. (2002). Schemas, Frames, and Scripts in Cognitive Psychology - International Encyclopedia of the Social & Behavioral Sciences. Retrieved from Goldman, A. I. (2015). Theory of human action. Princeton University Press.