Cognitive Information Processing Model
This article is small research about the Cognitive Information Processing Model and its relation with AI and Robotics.
Cognitive information processing (CIP) is most commonly referred to as “Information processing” (Schunk, 1996). It focuses on how people attend to the environmental events, encode information to be learned and relate it to the knowledge in memory, store new knowledge in memory and retrieve it when needed (Schunk, 1996). Two major contributors of CIP theory are Atkinson and Shiffrin who developed a multi-store model to explain the cognitive processes. CIP theorists compare the way our brain works to the functions of a computer (Schunk, 1996). This multi-store model composes of three stores which include: ‘Sensory’, ‘Short-term’, and ‘Long-term’ memories (Driscoll, 1994).
Sensory memory stores information linked with the senses, such as sight and smell, for enough time so that the information can be processed further while short-term memory (STM) decides where, when, and how to organize and store information (Driscoll, 1994). Short-term memory is a temporary memory where further processing of information takes place to make it ready for a response or to store it in the long-term memory. CIP theorists believe that the short-term memory capacity is limited to seven plus or minus two items of information, thus it is also termed as ‘working memory’ (Driscoll, 1994). Your short-term memory can hold just a finite amount of information and for a finite amount of time. Short-term memory comprises of three processes:
1. Rehearsal: The repeated assessment of information for transferring it to the long-term memory. This can occur through flashcards or studying the information over and over again. Elaboration during rehearsal can help solidify the information in the long-term memory (Driscoll, 1994).
2. Chunking: The process of breakdown large amounts of information into manageable parts to enlarge the capacity of short-term memory (Driscoll, 1994).
3. Coding: The process of converting information into a useable form such as mnemonics, outlines, and graphic organizers. Retrieval of information from LTM also occurs during coding to help the STM make sense of the new information (Driscoll, 1994).
Long-term memory (LTM) is our permanent store of information and it has the ability to store immense and diverse amounts of information. Information can be stored in LTM in a variety of ways including a network of nodes linked by paths, buckets of information that share similar characteristics, proposition statements linking a concept to a single characteristic, etc. (Driscoll, 1994). CIP theorists are also interested in the retrieval of information and forgetting of data stored in the LTM. The retrieval of information is initiated by the STM which searches for the information to make sense of the information in the holding cells. Cue in the environment can help in the retrieval of information. CIP theorists believe that information is permanent in LTM but forgetting can occur as a result of bad coding, interference, or inadequate cues that lead to failure in searching to retrieve the information (Driscoll, 1994) (Michael Beetz, 2007).
Role of CIP in Artificial Intelligence
The Cognitive Information Processing Model plays a significant role in Artificial Intelligence (AI) and is a hot area of research today in the field of AI. Cognitive capabilities are integrated into cognitive AI demonstrators and then researchers in engineering and computational science cooperate with each other to achieve the demonstration scenarios (Michael Beetz, 2007). The cognitive abilities of CIP modeled AI systems enable them to check results, debug and use a better approach if one fails. This ability makes them more intelligent than conventional AI systems (Michael Beetz, 2007).
Robots can do work much faster than a human but there are many daily routine tasks like driving which these robots are unable to do because they do not have a brain while on the other hand humans are much weaker but they can do those tasks easily because they possess a brain and an information processing model which is very flexible and adaptable; robots, however, cannot show such level of flexibility and adaptability (Michael Beetz, 2007). Robots are unable to do a task that is highly unpredictable and unreliable. Cognitive scientists are developing findings of neurobiological and neurocognitive cognition and developing a computational model (Michael Beetz, 2007). These models are evaluated and applied to CIP modeled AI systems.
Cognitive architectures have been introduced to achieve three main goals (Antonio Lieto, 2018). These include:
i. To record the invariant mechanisms of human cognition, those that include actions and functions such as learning, memory, reasoning, and perception, etc. (Antonio Lieto, 2018).
ii. To form a basis for the development of cognitive ability in a mechanism over a period of time, such as the factors that determine its cognitive initiation and response (Antonio Lieto, 2018).
iii. To attain human-level intellect, referred to as ‘General Artificial Intelligence’ by realizing artificial artifacts (Antonio Lieto, 2018).
The design of cognitive approaches for different systems have followed unique approaches obviously, but all work towards a unified goal of building towards a sentient human-like system. What better framework to use to build a humanoid robot than to use models inspired by actual human cognition? However, models such as the CIP are very restricted to their functionality (Antonio Lieto, 2018). While they have provided a strong basis, they cannot be directly used to solve challenges beyond single components, which then require a greats amount of architectural abstraction (Antonio Lieto, 2018).
Model Implications in light of Cognitive Robotics
The ultimate goal in the development of cognitive robotics according to Brachman is “ones that can reason using substantial amounts of appropriately represented knowledge, learn from its experience so that it performs better tomorrow than it did today, explain itself and be told what to do, be aware of its own capabilities and reflect on its own behavior, and respond robustly to surprise.” (Brachman, 2002). The cognitive systems have dynamic mechanisms of information processing rather than static coding that’s why they are reliable and ease communication and cooperation with humans (Brachman, 2002). A cognitive robot is to exhibit more than a “classical” robot, excelling beyond mainstream functions such as navigation, motion planning, etc., developing more semantic abilities, and enhancing its decision-making and context interpretation, allowing more efficient human-robot interaction (Rachid Alami, 2006).
Rapid advancements in the field of robotics and artificial intelligence have much potential to bring positive change in the lives of people but besides this, they may drastically change the lives of people by replacing them in simple jobs with robot workers. Moreover, cognitive robots due to their capability of learning and reasoning would even be able to take decisions even in complex situations, thus removing the human element in many scenarios and creating more concern over the implementations of such technologies in the industry (Brachman, 2002). That is why people who have knowledge about upcoming robots have positive expectations from such developments in their daily life, but their attitudes also show worry over the implications of such technologies for a whole society.
Cognitive robotic systems can not only play a major role in industries, and the economy of a country, but can cover a vast area of even helping old and disabled people or citizens in their homes through technologies such as car automation; as explored through a survey, by Pew Research Center in the United States (Aaron Smith, 2017). A practical example that we can analyze can be of a kitchen assistant which is developed to help people as their servants, enhance their cognitive capabilities and monitor their health and safety (Michael Beetz, 2007). To perform all these tasks the assistive kitchen has to learn, analyze and model all the house tasks and adapt itself according to the needs of the people. The robot has a camera for long-range object detection and an RF-ID reader for object identification (Michael Beetz, 2007). All kitchen accessories are smart and robots can access these smart accessories like refrigerator and microwave oven. The robot needs a lot of tasks related to its perception and cognitive learning. It has to determine the everyday tasks, tools needed for each task, and all respective arrangements. The robot has to perform many activities concurrently such as cooking and washing dishes and can even be interrupted if the doorbell rings or any person call him (Michael Beetz, 2007). So, it involves a high level of cognitive learning because each activity is combined by many small activities and robots take many steps to perform one activity.
Researches, Advancements, and Challenges
The field of cognitive robotics is a highly advancing field that is constantly upgrading and providing new challenges to researchers, developers, as well as philosophers, and psychologists. One particular exciting field of research comprises the achievement of consciousness in a cognitive robot. Consciousness is still a murky topic for researchers today, who have not really been able to understand the phenomenon, and philosophers wonder its implications in cognitive robotics would aid in understanding it better (James A. Reggia, 2018). Therefore, a study was conducted to show the reach of an imitating robot to limited consciousness conducted by the University of Maryland in early 2018. The team proposed to explore a possible consciousness framework for a cognitive bot that learned its tasks through the task of imitation (James A. Reggia, 2018). The team successfully managed to implement top-down gating of neurocomputational frameworks and top-down memory gating. They were able to successfully conclude that cognition and consciousness were strongly intertwined and could be achieved by vastly expanding their neural framework (James A. Reggia, 2018).
Another study proposes a different framework for a fully cognitive and interactive robot. Pacherie devises a three-level architecture consisting of the ‘Shared Distal level’ which deals with intention issues such as commitment, the comes a ‘Shared Proximal Level’, which comprises of the execution of the said planned action, and finally the ‘Coupled Monitor Intention’ which ensures real-time coordination of the execution action (Sandra Devin, 2016). Similarly, a number of researchers are trying to explore the depths of cognition in robots, plunging into subcomponents such as consciousness, socialism as well as emotions. The role of emotions in natural human and animal cognition has stirred a great rise in its contributions in the fields of computational models for artificially intelligent systems (Tom Ziemke, 2009).
However, with the advancements in technology, there may arise some ethical issues which must be kept in mind. The world has seen huge social complexity with the arrival of social robots, computer graphics, and virtual tools, etc. In order to tackle this problem scientists and engineers have come up with interdisciplinary research group activities in order to determine the rate of evolution with intelligent tools so that the ethical challenges which may arise in education, culture, gaming, nursing, and therapy may be resolved (Ali Meghdari, 2018).
There may be some technical and technological problems while developing cognitive robots that might occur due to insufficient programming skills and evolving behavior may result in unpredictable, risky, and dangerous robots. So, upgrading them to the new level of physical skills and cognitive capabilities is another great challenge because they must be according to the expectations of the human mind as the human-robot interaction is completely different from human-machine interaction (Ali Meghdari, 2018).
Several researches have been carried out on general-purpose robots that might be used in rehabilitation nursing and rescue operations. It has provided scientists with a new level of strength, robustness, cognitive ability, and intelligence. In order to fulfill the needs and expectations of the human mind with cognitive abilities, a new field has been evolved by the name Social Cognitive Robotics the main focus of which is a human-centered design of technological oriented systems (Ali Meghdari, 2018). Scientists are working on social interactions, designs, and working strategies to form a combination of human knowledge that can be used to update the system design. There are two main areas of social cognitive robotics: how people interact with current tools and technology and the other being the system design to build new interactive technologies. The main aim is to teach robots how to deal with complex situations. Similarly, various experiments have been made on people with different disorders and it has played a very positive role in social performance, eradication of distress, and in the enhancement of learning behaviors (Ali Meghdari, 2018).
Handling the social behavior of robots has become a problem for both humanities and robotics engineering because humanity is at the threshold of replicating an intelligent and autonomous system (Ali Meghdari, 2018).
In this regard, robotics play a vital role that governs the principle of morality like the distinction between good/bad and right/wrong. Similarly, there are some ethical laws that were given by Issac Assinov in 1950 which were.
· Robots may not injure a human being directly or may any harm may not be caused due to them.
· They must obey the orders given by human beings except those which contradict the first law.
· They must protect their existence if they do not conflict the first and the second law.
(Ali Meghdari, 2018)
One of the societal aspects that they may harm is the issue of economy and employment (Ali Meghdari, 2018). They may replace people in industries and at workplaces. Similarly, society would be so dependent on them just like its dependency on mobile phones and electronic devices (Ali Meghdari, 2018). Their affordance and ease of access would also be a disturbing factor for society. A small error in their development and coding would lead to fatal results like in smart cars and robots. Likewise, there is a struggle to make every part of the robot biological which may rise confidentiality and privacy issues and a time may come when they start to ask for their legal rights (Ali Meghdari, 2018). There may be involved in criminal activities due to which a special police force would be required for them.
Conclusion
The Cognitive Information Processing Model (CIP) is a critical contribution to the fields of both human and machine study. This framework, which essentially describes the fundamentals of how human cognition functions, is the basic key to enhancing robots and machines to achieve human-like behavior more and more. The goal of science and technology is to keep advancing and make systems as human-like as possible to ensure the most efficient human-robot communication. The CIP model is an integral part of the vast field of Artificial Intelligence today, that has revolutionized the robotic industry, as can be seen by creations like “Sophia”, a humanoid robot.
However, naturally there a lot of complications and challenges that are to be addressed in this quest, such as those of achieving consciousness and emotional responses, which are the center of research works lately. However, the contributions of the CIP model in achieving this huge industrial and academic success in the field of cognitive robotics cannot go undermined (Antonio Lieto, 2018).
References
Aaron Smith, M. A. (2017, October 4). Automation in Everyday Life. Retrieved from Pew Research Center: https://www.pewresearch.org/internet/2017/10/04/automation-in-everyday-life/
Ali Meghdari, M. A. (2018). Recent Advances in Social & Cognitive Robotics and Imminent Ethical Challenges. 10th International RAIS Conference on Social Sciences and Humanities (RAIS 2018). 211. Atlantis Press.
Antonio Lieto, M. B. (2018, May). The Role of Cognitive Architectures in General Artificial Intelligence. COGNITIVE SYSTEMS RESEARCH, 48, pp. 1–3.
Brachman, R. J. (2002, November). Systems That Know What They’re Doing. IEEE Intelligent Systems, 67–71.
Driscoll, M. P. (1994). Psychology of learning for instruction. Boston : Allyn and Bacon.
James A. Reggia, G. E. (2018, January 26). Humanoid Cognitive Robots that learn by Imitating: Implications for Consciousness Studies. frontiers in Robotics and AI.
Michael Beetz, M. B. (2007). Cognitive Technical Systems — What Is the Role of Artificial Intelligence? 30th Annual German Conference onAI, (pp. 19–42). Munich.
Rachid Alami, R. C. (2006, May). Towards Human-Aware Cognitive Robots.
Sandra Devin, G. M. (2016, March 2). Some essential skills and their combination in an architecture for a cognitive and interactive robot.
Schunk, D. H. (1996). Learning Theories An Educational Perspective (Vol. 6th). Pearson.
Tom Ziemke, R. L. (2009, February 6). On the Role of Emotion in Embodied Cognitive Architectures: From Organisms to Robots.
