Artificial General Intelligence

Prediction and Explanation

1. Representing time

Events are statements with time-dependent truth-values.

Temporal relations are defined relatively between events. There are two basic temporal relation: "before--after" and "when". Here NAL allows different granularity and accuracy in representation.

Event-related terms and concepts have time-dependent meaning.

To "reason about time" and to "reason in time" --- reasoning takes. NARS gets a "personal sense of time" using its internal working cycle as a clock.

The "real-time experience" of NARS measures the time interval between perceived events.

Temporal relations are combined with logical connectors and copulas: two versions of conjunctions and 3 versions of implication (equivalence).

Tense: event compared to "now". Problem: "now" is a moving reference. Tense is used only in external communication, not in internal representation. The need to attach a time-stamp to each sentence to remember is creation (perceiving or deriving time).

Multiple ways to represent temporal information: (1) built-in temporal relations, (2) internal clock and tense, (3) acquired temporal relations.

2. Temporal inference

In IL-7, the logical and temporal relations are processed in parallel by the inference rules. Then the inference rules are extended into NAL-7 strong rules, and their reversions becomes NAL-7 weak rules.

Tensed sentences are processed similarly.

Temporal inference is used to explain the past and to predict the future.

Temporal induction and classical conditioning.

Causal relation as an acquired notion and its domain dependency. There are also different types of explanation.

"Meta-level induction": to use tensed judgment as evidence for eternal judgment.

3. Reasoning as a unified cognitive mechanism

Multiple functions of question-answering: The relation between functions and inference types is not one-to-one, but many-to-many.

Factors influencing the result from the choice rule: truth-value, simplicity, priority-value.

Also, as proposed by Peirce:

However, in general a cognitive function is provided in multiple ways by the cooperation of various inference rules.

Compared with the various cognitive architectures, the unified approach followed by NARS has the advantages of consistency and simplicity.

4. Knowledge self-organization

The system constantly reorganizes its knowledge for question answering.

The major criteria in organizing knowledge:

The processes of axiomatization and deaxiomatization coexist.

An object or process often can be described at different levels of description, with different granularity and scope. These levels are related, but cannot replace each other.

The relation between levels of description is different from the relation between an object-level and its meta-level.


Reading