There seems to be a few things to consider.
The first thing to know is that search engines do not match keywords as the SEO community would like you to believe. Instead, various forms of linguistic semantic analysis is conducted against the content. Here is more of how that works. Please stick with me, it will all make sense in the end.
When a web page is analyzed, the HTML DOM model is broken into sections based upon elements such as header tags, paragraphs, tables etc. To keep things simple, let us assume the content is headers and paragraphs. There may be images and links to resources etc., but I wan to focus on the mechanics in a simple way.
Each element is assigned an ID in the order that they are found as well as groups of elements. Paragraphs between headers are a single block of content that belongs to the header above it. Each paragraph is also a block belonging to the larger block and header tag with a value within the ID that allows the paragraphs to ordered. Each set of paragraphs and individual paragraphs are given an ID that makes it a child of the header so that references between the blocks and the header are clear and manageable.
Each header, individual blocks, and sets of blocks goes through several forms of semantic analysis to determine weighting of the block. This analysis exists in the form of a matrix (think of a small spreadsheet style table). The scores within the matrices are then used to match intent in a variety of ways including expertise, reading grade level, topic, term weight, citations, etc.
After the SERPs are created, then search query terms are highlighted. This gives the impression that the term matches are made when in fact, a page can rank for terms and topics that are not always represented within the content simply because there is a match in an ontology somewhere.
When a query is made and you examine the SERP results, you will not see header tags within the SERP snippet, but paragraphs and other forms of content will appear. This is important to know. This does not mean that header tags are not important. They are. It just means that content is of more value in some manner because content gives a fuller picture as to what the content is about. In that respect, header tags are weighted as important elements and content presented to the user as being relevant in a more human form.
The most valuable HTML elements in semantic analysis (in relative order) is the title tag, h1 tag, description meta-tag, internal and external inbound (back) links, etc. While header tags are analyzed, it seems that the h1 tag carries a high importance for the page overall since it defines the topic of the page along with the title tag, and any other header tag is of high importance for the content blocks following.
Both of your examples have issues.
In your first example, the paragraph tags with brand name and product name carry a narrow and highly focused semantic value. For example, for a brand such as Sony, the paragraph tag does not answer the question: What about Sony? What are you trying to say? This would be the same for a header by the way.
Example 1 has single terms with little semantic value as content that would get lost and not offer much value.
Example 2 has single terms in a header tag that defines the content to follow. While it is just a single term, these terms are very specific and focused.
With example 1, it would be expected that the contents of the paragraph would contain a subject, predicate, and object like a normal sentence would, whereas with example 2, it could go either way - single term or conversational.
If, for example, you want semantic value to the terms as you are using, and you have them represented in the title tag, h1 tag, and description meta-tag, and it is not important that you necessarily have this portion of content appear in the SERP snippet, then a header tag is the best option. For example, you may have the brand name and product name in the product description and that would help the user more and make a more valuable SERP snippet. The header tag would give semantic weight without appearing to directly match as content thus allowing other content to be used for the SERP snippet. As a paragraph, the opposite could happen and that portion of content (header tags are not really content per se' in matching search intent) could give a confused SERP snippet.
The next question is this. Are h4, h5, and h6 header tags of any value for search and SEO?
The answer is yes. Absolutely. More than people realize. But maybe not the way people realize. I will explain.
Some argue that there is a decreasing level of value for header tags. I am one of those people, however, the truth is that no-one except a search engine engineer can tell you how that is specifically structured. Some argue that the weight given to lower header tags is small and they would be absolutely right. But if someone was to argue that there is no effect, I will argue strongly that is NOT the case. Lower header tags can have significant weight under certain and not very limited circumstances.
In my research, the semantic value of 3 or 4 h6 tags can out perform an h1 tag but a wide margin. If someone is not careful, lower header tags can really skew the performance of a page. And this is precisely the point. Header tags are to represent the content as we know. Title tags along with the h1 tag generally represents what a page is about and lower header tags follow a hierarchical organization of the content blocks and what they are about. That is a given. If you were to consider that there is just one h1 tag and several, if not quite a few header tags below the h1 tag, which set of header tags offer search engines the best opportunity to really know what a page is about? The relatively short h1 tag that already closely matches the title tag or all of the likely more verbose header tags that follow? The answer is the latter. So while one individual lower header tag in of itself will not seem to make much of a difference, several lower header tags including the lowest, h6, can really significantly effect how a page is found in search when taken as a group.
To say that lower header tags have little or no effect in of itself is misleading without taking into account how header tags are used by the content author and search engines as a rule. Sure one lower header tag might not make a difference that is easily measured, but several will.
Here is one more point. HTML header tags and content blocks are semantically weighted using more than one algorithm and sometimes two (often) or more (rarely) in one analysis. They scores of the analysis are represented in a matrix and matrices (more than one matrix represented in a single matrix) for weighting and further analysis. The reason for this is simple. Each algorithm by itself does not paint a full picture. For content blocks and sets of content blocks to properly be understood semantically, the header tag will add weight to the following content blocks and sets of content blocks and not be weighed by itself in a vacuum. It is really not the header tag by itself that influences, but the added weight from the header to the content blocks that follow that influences.
